Introduction

Cotton (Gossypium hirsutum L.) production in the semi-arid eastern Mediterranean region, including northern Syria, is of concern because of the limited water resources and overexploited aquifers. In Syria, rapid increases in irrigated land (majority as ground water irrigation) and inadequacies in agricultural water delivery, use, and management practices contribute most to the national imbalance in water supply and demand (Varela-Ortega and Sagardoy 2003; Salman 2004; Jamal et al. 2006). Irrigation applications are nearly twice as high as cotton water requirements with a national average of 15,000 m3 ha−1 per season. Nearly three-fourths of cotton fields are irrigated by flooding of small basins. Because of over-irrigation, water productivity in the traditional surface-irrigated lands is about 0.2–0.25 kg m−3 of seed cotton per applied water (or 0.07–0.09 kg m−3 for lint cotton). These values are two to three times lower than most cotton producing regions in the world, such as Argentina, Turkey, and USA (see e.g., review in Grismer 2002 of studies by Hunsaker et al. 1998; Ayars et al. 1999; and others).

Because of the need to sustain agricultural production, the Syrian government adopted a national irrigation modernization plan in 2001, which encourages farmers to change to modern irrigation techniques (like drip and sprinkler irrigation) by providing technical support and tax-free low-interest loans. Although the adoption of modern irrigation systems is slow (i.e., national data show less than 5% of cotton land as drip irrigated), there is a great desire among Syrian cotton growers to convert to drip irrigation because of its potential to conserve water and reduce pumping cost, labour, and other inputs (Janat 2004). These benefits have been demonstrated in Syria, with research showing 35 to about 100% irrigation savings using drip and up to 50% increase in yield using drip with fertigation as compared to traditional surface irrigated cotton (Al-Darir 1998; Janat and Somi 2001). Besides improving irrigation system efficiency, there is a need for effective on-farm water management, i.e., proper scheduling of irrigation (Farahani et al. 2006). The most fundamental requirement of scheduling is the determination of crop evapotranspiration (ETc).

A popular method used to estimate ETc is the crop coefficient (K c) algorithm (e.g., in the UN’s Food and Agriculture Organization (FAO) article-24 by Doorenbos and Pruitt 1977). In this approach, an experimentally determined coefficient (K c) is multiplied by evapotranspiration from reference vegetation (ETo) to compute ETc, or

$$ {\text{ET}}_{{\text{c}}} = K_{{\text{c}}} \times {\text{ET}}_{{\text{o}}} $$
(1)

Doorenbos and Pruitt (1977) recommended the use of accepted equations to compute ETo (defined for short green grass with ample supply of soil water). More recently, the FAO-56 (Allen et al. 1998) promotes the more superior Penman-Monteith (P-M) combination equation. Tables of K c values derived from field and lysimeter ETc measurements are provided in literature (see e.g., Doorenbos and Pruitt 1977; Wright 1979, 1981, 1982; Jensen et al. 1990; Martin and Gilley 1993).

The practical simplicity of using the K c approach is indisputable, but the adoption of generalized K c curves can lead to errors (Hunsaker et al. 2003). Since local development of K c is a difficult task, most practitioners rely on the published values. This may be unwise because of the empirical nature of K c that may limit its transferability into locations where the local climate and management factors deviate from the conditions in which the tabulated value was developed. The K c approach is advocated in Syria, but no K c curve has been locally developed, and the performance of the generalized K c values is unknown.

The objective of this study is to develop the K c curve for drip-irrigated cotton in northern Syria using detail soil, plant, and climate measurements for the period 2004–2006. The locally developed K c curves are compared with generalized FAO K c values adjusted for local climate and management. The possible sources of errors in using the published K c and the difficulties in developing local K c curves are discussed.

Materials and methods

Site description and soil characteristics

This study was done in cooperation with Syria’s Cotton Research Administration as part of a broader experiment examining the effects of varying soil water regimes on water use efficiency of irrigated cotton (Gossypium hirsutum L.) in the cotton-wheat rotation system over the period 2004–2006. This paper presents data from the fully irrigated plots only. The study was conducted at Tel Hadya station (36° 01′N, 36° 56′E, and 284 m above mean sea level), the research headquarters of the International Center for Agricultural Research in the Dry Areas (ICARDA), located 35 km south of Aleppo, in northern Syria. The prevailing eastern Mediterranean climate in northern Syria is characterized with a single rainy season (Kassam 1981), starting in the fall and extending through the spring, during which the mean rainfall is 351 mm. There is limited rainfall in May and June and almost none in summer when irrigation is needed for a profitable production.

Soil at Tel Hadya is generally deep, 1.5–2.0 m, classified as fine clay (montmorillontic, thermic, Chromic Calcixerert), with volumetric water contents at field capacity and wilting point determined at 38 and 22%, respectively (Ryan et al. 1997).

Crop practices

The short season cotton variety ‘Aleppo-118’ was similarly managed over the 3-year study. This variety is characterized by high productivity, good gin average, and tolerance to Verticillium wilt. The experimental plots (10 m wide by 13.3 m long) were sown by hand, with seeds spaced 0.2 m apart on 0.7 m flat rows. Table 1 summarizes the planting and harvest dates for the 3 years. There were three rates of nitrogen (N) applications, 200, 150, and 100 kg N ha−1. Each nitrogen treatment was replicated three times, for a total of nine plots. Nitrogen application (46% N urea) was split, with 1/5th applied at planting and the rest fertigated on three occasions during each season. Adequate phosphorus (P2O5) was broadcasted prior to sowing and based on pre-season soil testing. The field was monitored for insects and diseases and chemicals were applied as needed. Weed control was achieved using manual and chemical methods. Seed cotton was harvested by hand on two occasions (Table 1), in order to duplicate local farm practices, with the first picking producing more than 70% of the total yield. For analysis, the period from 6 May (one day after emergence, DAE) to the second picking on 24 September (142 DAE) was selected as the best representation of the growing season in the 3 years, providing a fixed duration for comparative purposes.

Table 1 Summary of planting and harvest dates for the 2004–2006 growing seasons
Full size table

Irrigation management

The replicated plots were irrigated to meet full crop water requirements using well water applied through a drip irrigation system. The irrigation system was designed and managed to ensure uniform application. Polyethylene drip laterals (16 mm inside diameter) were installed after sowing in every plot (except in 2004 where the drip system was made ready after the first irrigation with a mobile irrigation system), laid along every crop row with emitters (rated at 4 L h−1 discharge) spaced every 0.4 m on the laterals, and operated at plot-inlet pressures between 110 and 150 kPa. Observations of emitter wetting patterns after irrigations showed complete closure between adjacent emitters on the same lateral and more than 85% closure between the rows. The applied water per plot was measured using mechanical flow meters installed at plot-inlets, occasionally verified against multiple measurements of emitter flow rates.

Soil water content was periodically monitored using neutron scattering. Prior to sowing, an aluminum access tube was installed in the center of each plot to a depth of 1.80 m. Soil water content was measured using an on-site calibrated neutron probe (Type IH-II, Didcot Instruments, Co, Ltd., Abington, UK) at a minimum of weekly intervals, for 30–35 readings per season. Measurements were made for each 0.15 m layer in the soil profile to the bottom of access tube, except the top 0.15 m, which was measured gravimetrically. Irrigation was initiated when soil water in the top three to four soil layers approached, but never dropped below, 50% of available soil water (difference between field capacity and wilting point water contents) and refilling the profile to field capacity. The first irrigation was applied after sowing, the second irrigation about 35 days after, while the irrigation season ended by late August, allowing late-season soil water drawdown to expedite boll opening (Table 2).

Table 2 Drip irrigation scheduling (frequency and amounts) for the 2004 to 2006 growing seasons
Full size table

FAO-56 Kc curve

The general shape of the K c curve (shown later in Fig. 5) resembles the changes in the vegetation and ground cover during plant development and maturation that affect the ratio of ETc to ETo. FAO-56 identifies the K c curve by the following four crop growth stages:

  • Initial-season: Period from planting to approximately 10% ground cover,

  • Crop development: Period from 10% ground cover to effective full cover or peak water use (which ever comes first),

  • Mid-season: Period from effective full cover to the start of maturity, with the latter often marked by the beginning of ageing, yellowing or senescence of leaves, and leaf drop associated with decline in water use, and,

  • Late-season: Period from the start of maturity to harvest or end of water use.

The K c curve is constructed knowing the duration of each growth stage in addition to K c values for the initial stage (K c-ini), the mid-season stage (K c-mid), and at the time of harvest (K c-end). Determination of crop growth stages and their subsequent matching with the four FAO-defined growth stages is not straightforward for indeterminate species such as cotton. It is also noted that cotton top growth continues if the season is not terminated by soil water drawdown or chemical application to induce defoliation or desiccation. This latter method is common to many cotton-producing regions worldwide, but not in Syria. The duration of each crop growth stage was determined from field observations and varied by up to 5 days in the 3 years, averaging 39, 34, 30, and 39 days for the length of initial, development, mid-season, and late-season growth stages (Table 3).

Table 3 Calendar and cumulative heat unit (Growing Degree Day, GDD) periods for the four cotton growth stages
Full size table

In FAO-56, the generalized K c values for cotton are K c-ini = 0.35; K c-mid = 1.15–1.20 (average = 1.17); and K c-end = 0.70 to 0.50 (average = 0.60). FAO-56 suggests various adjustments of these K c values caused by local variations in climate, soil and irrigation management. The K c-ini values are generally below 0.4 for average wetting frequencies, but may approach values as high as 1.0–1.2 (FAO-56) for high frequency irrigations. For the experimental site, FAO-56 recommends (page 118 for fine textured soil) adjusting K c-ini to a value of 0.2, because of high evaporative demand (average ETo for May was 7.4 mm per day) and infrequent wetting events.

The FAO-56 K c values are expected in a sub-humid climate having average daily minimum relative humidity (RHmin) values of about 45% and calm to moderate wind speed (u 2) averaging 2 m s−1. At the experimental site, average RHmin and u 2 during the mid-season were 31% and 5.3 m s−1 (2004), 30% and 5.2 m s−1 (2005), and 27% and 5.5 m s−1 (2006), respectively. To account for the lower humidity and higher wind conditions, the K c-mid value of 1.17 was corrected for each year using the FAO-56 recommended correction formulae

$$ K_{{{\text{c-mid}}}} = K_{{{\text{c-mid}}\;{\text{unadjusted}}}} + {\left[ {0.04(u_{2} - 2) - 0.004{\left( {{\text{RH}}_{{\min }} - 45} \right)}} \right]}(h/3)^{{0.3}} $$
(2)

with measured mean values for RHmin, u 2, and h (crop height, m). A similar adjustment was made for K c-end. The measured crop height varied between 0.75 and 0.80 m during these two growth periods. The FAO recommended adjustments increased K c values for the mid- and end-season, reflecting the hot and dry climate conditions (Table 4).

Table 4 Generalized FAO K c values adjusted for local climate and the locally developed K c values for cotton
Full size table

Cotton evapotranspiration

Soil water budget method was used to estimate crop evapotranspiration. This involved measuring the components of the water balance equation for a control volume defined by soil profile of given root zone depth and written as:

$$ {\text{ET}}_{{{\text{c-Meas}}}} = P + I - D - R - \Updelta S $$
(3)

where ETc-Meas denotes measured crop evapotranspiration, P is precipitation, I is irrigation, D is deep percolation below the root zone (or an upward flow, if negative, into the root zone in case of a shallow water table), R is runoff (or run-on, if negative, in case of surface flow into the area under consideration), and ΔS is the change in soil profile water storage (the period ending minus the period beginning soil profile water), with all variables in units of equivalent mm water and determined for each period between two consecutive soil water measurement days. The reliability of ETc-Meas estimates from the soil water budget method depends on the measurement or estimation accuracy of the variables in the right-hand side of the equation.

Deep percolation (or upward flow) is the most difficult variable to quantify and accounts for most mass balance errors in estimating ETc-Meas. No shallow water tables existed at the study site, thus only percolation below the root zone was of concern. One source of mass balance errors is when the depth of the profile measured by the neutron probe is less than the wetting front by irrigation, with deep percolation, if any, undetected through the bottom of the profile (Wright 1990). In this study, the access tubes were installed up to 1.80 m, to provide sufficient depth for detection of potential deep percolation. An analysis of the soil profile water content measurements revealed negligible changes in soil water in layers below 1.20–1.35 m depth (Fig. 1 for 2006), suggesting limited percolation. Profile data also show that over 95% of root water extraction was from soil layers above the 1.20 m depth. For analysis, ETc-Meas for cotton was thus calculated from Eq. 3 using the entire 1.80 m profile for each period between two consecutive soil water measurements. Measured precipitation was less than 5 mm during summer in each of the 3 years, and thus no appraisal of its effective portion was made. The drip-irrigation system produced no runoff, with R equaling zero for analysis.

Fig. 1
figure 1

Profile soil water content at planting (4 May) and harvest (24 Sep) and at 1 day before and 2 days after the second (12 Jun), fifth (20 Jul), and ninth (24 Aug) irrigation events in 2006. (2B, 5B, and 9B denote measurement at one day before, and 2A, 5A, and 9A denote measurement at 2 days after, the second, fifth, and ninth irrigation events)

Full size image

Meteorological data and reference evapotranspiration

Daily values of precipitation, minimum and maximum air temperature and relative humidity, wind speed at 2 m height, and solar radiation were measured at an automated weather station inside the research center with a grass surface. Using the on-site climate data, daily ETo was computed with the Penman-Monteith equation as described in FAO-56 (Fig. 2).

Fig. 2
figure 2

Calculated daily reference evapotranspiration (ETo) at Tel Hadya experimental site for the period 2004–2006

Full size image

Locally developed Kc curve

Crop coefficient curves were locally developed for each of the three seasons by first determining average K c-ini and K c-mid values as the ratio of cumulated ETc-Meas to cumulated ETo for the initial- and mid-season growth stage periods, respectively. The K c-end was then determined by maintaining an ETc-Meas mass balance (for the late-season period) while calculating the slope of the descending K c. The locally developed K c values are referred to as K c-ini Local, K c-mid Local, and K c-end Local (Table 4) to be distinguished from the previously determined adjusted FAO K c values (i.e., K c-ini FAO, K c-mid FAO, and K c-end FAO).

Results and discussion

Irrigation and evapotranspiration

Seasonal irrigation amounts were 800, 810, and 760 mm, while mean seasonal ETc-Meas values were 895, 927, and 813 mm in 2004, 2005, and 2006, respectively (Table 5). Variability in the ETc-Meas values across plots was small each year (see Fig. 3 for 2004 data), with standard deviation ranging narrowly between 14 and 17 mm. Mean seasonal ETc-Meas was on the average 88 mm higher than the seasonal irrigation, reflecting the seasonal contribution of soil water. Compared to literature, the 3-year average ETc-Meas value of 878 mm in northern Syria is higher by about 90 mm than those reported for the medium maturity cotton in northern High Plains of Texas (Howell et al. 2004) and near Menemen, in western Turkey (Allen 2000), even though the season is shorter (by about 20 days) and the K c-mid value of 1.05 is smaller than their reported range of 1.15–1.30.

Table 5 Seasonal applied irrigation water, measured cotton evapotranspiration (ETc-Meas), and calculated Penman-Monteith reference evapotranspiration (ETo) in northern Syria
Full size table
Fig. 3
figure 3

Cumulated measured ETc for replicated plots of the same N application (top) and for plots with different N applications (bottom) in 2004

Full size image

Seasonal ETo values were 1204 and 1224 mm in 2004 and 2005, respectively, but increased to 1336 mm in 2006. The higher demands were caused by higher daily wind speed (by an average of 0.5–1.1 m s−1), and lower relative humidity (by 4–6%), during mid-season in 2006 as compared to 2004 and 2005. In spite of the higher ETo in 2006, measured ETc was lower by 82–114 mm in 2006 as compared to 2004 and 2005, respectively. More than half of this difference occurred during the late-season stage (the period 104 to 142 DAE), mainly because of wetter soil environments in 2004 and 2005 than in 2006. This is illustrated in Fig. 4 showing volumetric water content in the top 0.3 m soil for the three seasons. It is noted that there were three irrigation events during the late-season stage in 2005 for a total of 270 mm as compared to two irrigations in 2004 and 2006 for a total of 160 mm in each year. This plus the fact that the last irrigation was applied later in 2004 (29 Aug) and 2005 (2 Sep) than in 2006 (24 Aug) created wetter soil environments in the former years for enhanced late-season evaporation. The late-season variations in soil wetness had a significant effect on ETc-Meas and the locally developed K c-end Local value. For instance, ETc-Meas during the late-season stage was 306 mm in 2005, but a smaller value of 223 mm in 2006.

Fig. 4
figure 4

Volumetric soil water content in the top 0.3 m soil during the 2004–2006 cotton-growing seasons. (the three vertical lines mark the beginning of development, mid-season, and late-season growth stages)

Full size image

FAO versus the locally developed Kc curves

The ratio of ETc-Meas to ETo versus time defines the seasonal trend of locally developed K c (shown in Fig. 5 for 2006), where the spikes are due to high rates of evaporation from wet soil following irrigation events. These locally developed K c curves are compared to the adjusted FAO curves in Fig. 6 for the 3 years. The adjusted FAO curves were basically the same for the 3 years, with K c-ini FAO = 0.20, K c-mid FAO = 1.30, and K c-end FAO = 0.71. The locally developed K c curves not only differed considerably among the 3 years, but were also different than the adjusted FAO K c values. Deviations between the adjusted FAO and locally developed K c values ranged between −47 and 103%, with the widest deviations occurring for the initial- (ranging from −47 to 1%) and late- (ranging from −25 to 103%) season stages. For the mid-season stage, the locally developed K c-mid values were very similar (1.04 and 1.05), but lower by about 24% than the adjusted FAO K c-mid value of 1.30. Hunsaker (1999) developed K c for short season cotton in Arizona finding higher K c values than the ones proposed in FAO-56. Similarly, cotton K c values were up to 35% higher in Arizona and California than those reported in FAO-56 (Grismer 2002). In contrast, the locally developed K c values in this study are generally lower than the adjusted FAO values, i.e., by 24% for mid-season stage. The K c-end Local values found in this study are generally high (except in 2006), but comparable to those reported by Grismer (2002) with values of 0.87 for Sacramento and San Joaquin valleys and 0.95 for California desert counties. When transpiration is terminated by top-kill practices, the K c-end is substantially reduced, e.g., to values of 0.1–0.2 (Howell et al. 2004).

Fig. 5
figure 5

Seasonal trend of K c calculated as the ratio of ETc-Meas to ETo for each period between two consecutive soil water measurement days during the 2006 cotton-growing season (symbols mark the beginning of each soil water measurement period). (The thick grey line is the average K c for each growth stage period.) (Vertical lines show irrigation applications, 1 through 9, and amounts.)

Full size image
Fig. 6
figure 6

Locally developed K c curves for the 2004–2006 growing seasons (top), and their average curve compared to adjusted FAO K c curve (bottom)

Full size image

The results show considerable level of sensitivity of the K c methodology to normal year-to-year irrigation management variations during the early sparse canopy conditions and late season senescence. For instance, the average daily ETc-Meas and volumetric surface water content (top 0.3 m soil) values for the late-season period (Table 6) were highest in 2005 (8.1 mm d−1 and 26.1%) and lowest in 2006 (5.9 mm d−1 and 22.6%). The consequence of this is reflected in the calculated K c-end Local that was also highest (0.95) in 2005 and lowest (0.35) in 2006, with an intermediate value (0.67) in 2004. The results translate to an average 17% increase in K c-end Local value for each percent increase in topsoil water content. This sensitivity is known in literature, i.e., previously addressed in Jensen et al. (1990) and Allen et al. (1998), and suggests the use of multiple years of data to construct a more representative K c curve. The average of the three locally developed K c curves (Fig. 6) is thus preferred for scheduling of drip-irrigated cotton in northern Syria. Since most farming practices in Syria are suboptimal, the above-suggested average K c values may need to be lowered for field applications. For instance, K c-mid values developed originally under pristine cropping conditions required a 6% reduction in southern California (Allen et al. 2005) and 15% in western Turkey (Allen 2000) because the assumed optimal conditions did not prevail.

Table 6 Summary of measured cotton evapotranspiration and topsoil water content for the late-season growth stage in 2004–2006
Full size table

Crop coefficient as a function of thermal units (i.e., Growing-Degree-Days, GDD) has been suggested to improve accounting for climate variability and enhancing transferability of K c curves (Sammis et al. 1985; Slack et al. 1996; Hunsaker 1999; Howell et al. 2004). Cumulative GDD values were computed (Table 3), suggesting the required 380, 430, 410, and 450 degree days (°C) for the initial-, development-, mid-, and late-season growth stages. The average cumulative GDD for the cotton season was 1680°C in northern Syria, a value similar to data from the semi-arid New Mexico, USA (Sammis et al. 1985), but higher than those in northern Texas (Howell et al. 2004).

Predicted and measured crop evapotranspiration

The performance of the adjusted FAO K c curves is evaluated by using Eq. 1 to predict ETc (referred to as ETc-FAO) and compared with measured ETc (i.e., ETc-Meas) values. The comparison is presented in Fig. 7, showing that use of the adjusted FAO K c overestimated ETc in all 3 years and during all growth stages (except in the initial stages in 2004 and 2005). On the average, ETc-FAO per growth stage underestimated the measured values by 30% for the initial stage and overestimated by 23, 24, and 19% for the development, mid-season, and late-season growth stages, respectively. It is noted that the highest measured ETc occurred during the mid-season stage for a total of 320 mm and the lowest during the initial stage for a total of 87 mm.

Fig. 7
figure 7

Measured and predicted (multiplying adjusted FAO K c by ETo) cotton evapotranspiration during the 2004–2006 growing seasons. (Measured values are the result of multiplying locally developed K c by ETo.)

Full size image

As pointed out by Farahani et al. (2007) “How precisely does ETc need to be since irrigation application (depth or volume) and inherently field soil and crop variability can be much greater than ETc errors?” With care, using an appropriate model for ETo and reliable K c generally produces estimates of ETc within the accuracy of most field-irrigation systems to deliver water (Jensen et al. 1990). In this study and on the seasonal basis, ETc-FAO overestimated ETc-Meas by 10, 10, and 33% in 2004, 2005, and 2006, respectively, for an average of 17% (or 150 mm water) for the 3 years. Considering that the average irrigation amount per event was 85 mm in the 3 years, the overestimation of ETc by using the adjusted FAO K c curve is substantial and equivalent to about two additional irrigations per season.

Conclusions

The development of K c curve for cotton was pursued because of its simplicity, wider appeal to extension personnel, and limited data requirements for irrigation scheduling and water management. Two sets of K c curves were developed, the generalized K c values published by FAO that were adjusted for local climate, and the locally developed K c curves as the ratio of measured ETc to ETo for the 3 years. The locally developed K c curves not only differed among the 3 years, but also from the adjusted FAO K c values. The initial- and end-stage K c values appear to be the most susceptible to local variations than the mid-season value, presumably because of lower canopy cover and higher soil evaporation following wetting. Because of this, use of multi-year data is suggested for K c development. The use of the adjusted FAO K c values overestimated seasonal crop evapotranspiration by 10–33%, thus cautioning against their blind application without some verification.

Since neither the in-season soil water measurements, nor the use of computerized irrigation scheduling programs are yet adopted by Syrian farmers, the help of local extension service in an effective irrigation scheduling is needed. A practical approach might require a dedicated effort from extension to measure, compute, and publicize daily or weekly crop water requirements for the main crops in the area. The K c curves developed herein can enhance that effort. It is noted that the experiment reported herein was conducted at ICARDA experiment station, which resembles optimal (i.e., pristine) cropping conditions as compared to many farming practices in the region. For on-farm practices under less than optimal conditions, lower K c values than those reported herein are most likely required. Similar experiments under farming conditions are needed for that purpose.