3.1. Meteorological Conditions
Figure 3 shows the hourly net radiation (
Rn) at the Yingke Site from November 2007 to January 2009; these data reflect the seasonal characteristics of the
Rn. The maximum
Rn of each season was 649.28 (spring), 777.93 (summer), 663.54 (autumn), and 473.36 w∙m
−2 (winter). The maximum monthly total
Rn was 409.90 MJ∙m
−2 (June 2008), and the minimum monthly total
Rn was 23.35 MJ∙m
−2 (December 2008). The total
Rn was 2547.73 MJ∙m
−2 in 2008 (
Table 2). Temperatures are influenced by the
Rn. The air temperature (
Ta) was observed 2 m above the ground surface, and the annual average value was 7.07 °C in 2008. The monthly average
Ta reached a maximum (20.59 °C) in July and a minimum (−14.03 °C) in January (
Table 3). The hourly average
Ta was below 31 °C in spring, rose to a maximum of approximately 34 °C in summer, then fell below 28 °C in autumn, and descended to a minimum below 22 °C in winter. Except in January, February and August in 2008, the monthly and annual averages of
Ta were lower than the values from 1951 to 2000, as shown in
Table 1.
The monthly average wind velocity observed 2 m above the ground surface generally ranged between 1.0 and 2.0 m∙s
−1 during the observation period (
Table 3). The maximum hourly average wind velocity in 2008 was 9.35 m∙s
−1. The average wind velocity in 2008 was 1.23 m∙s
−1, which was lower than the average wind velocity from 1951 to 2000, as listed in
Table 1.
The monthly average specific humidity observed 2 m above the ground surface reached a maximum (7.23 g∙kg
−1) in July and a minimum (−1.07 g∙kg
−1) in January (
Table 3). The specific humidity was below 10 g∙kg
−1 in spring, rose to a maximum of approximately 20 g∙kg
−1 in summer, then fell below 15 g∙kg
−1 in autumn and descended to a minimum below 5 g∙kg
−1 in winter. The average specific humidity in 2008 was 3.66 g∙kg
−1. Except in November and December 2008, the monthly and annual averages of the specific humidity were lower than the average specific humidity from 1951 to 2000 listed in
Table 1.
Seasonal variations of rainfall are shown in
Figure 4. The total rainfall in 2008 was approximately 117 mm, which was lower than the values from 1951 to 2000 (
Table 1); the rainfall was concentrated (92% of the annual total) in summer and autumn (June to November 2008). During the observation period, the maximum monthly total rainfall was 37.20 mm (June 2008,
Table 2). Rainfall did not occur in November and December 2008. The seasonal distribution of rainfall was 8.00 (spring), 64.50 (summer), 43.70 (autumn) and 0.70 mm (winter). The total rainfall was approximately 70 mm during the entire maize-growing stage (from 20 April to 22 September 2008).
The soil froze in winter, and the underlying land was seasonally frozen ground (
Figure 5a). The maximum depth of the frost penetration reached up to approximately 100 cm. The average annual ground surface temperature (
Tg) in 2008 was 7.00 °C. The seasonal variations in the soil moisture content are shown in
Figure 5b. In winter, the soil moisture content was lower because of ground freezing. After March, the ground thawed, and the soil moisture content increased. Extreme soil moisture contents occurred after each irrigation event, which led to the high value centres at 20 cm. Generally, there was a high value belt at a depth of 120 cm. Vertically, the soil moisture content was lowest at 10 cm and highest at 120 cm (
Table 4). As shown in
Figure 5b, the soil moisture content peaked during each irrigation event except for the irrigation event on 25 August, because the soil moisture content data were missing. The monthly total soil heat flux ranged from negative to positive in March, and reached a maximum in May, a trend that was similar to that of the
Tg. The total soil heat flux decreased to negative values again in July (
Table 2) due to the oasis “wet island” effect [
16,
51], which indicates that the high amount of latent heat flux results in a cold land surface and decreases the sensible heat flux and the soil heat flux, even to negative values.
3.2. Seasonal Variations of Actual Evapotranspiration (ETa)
The seasonal variations in the
ETa at the Yingke site from November 2007 to January 2009 are shown in
Figure 6. The daily mean
ETa was 1.49 in spring, 3.90 in summer, 1.41 in autumn and 0.22 mm∙day
−1 in winter. In 2008, the total
ETa was 654.69 mm, and the daily average
ETa was 1.79 mm. The
ETa observed in the Zhangye oasis cornfield was higher than the
ETa observed in an arid oasis ecosystem of the Syrian desert in Palmyra [
52] and a
Tamarix ramosissima ecosystem in the extremely arid region of northwestern China [
21].
In 2008, the emergence time of maize occurred on 6 May, and the shooting stage of maize began on 19 June. The heading stage of maize began on 20 July. The filling stage of maize occurred from 5 August to 10 September, and the maturity stage occurred from 11 September to 22 September. The crops were harvested on 22 September at the observation field [
53]. During the entire maize-growing stage (from 20 April to 22 September 2008), the total
ETa was approximately 500 mm with a daily average
ETa of 3.33 mm∙day
−1. As shown in
Table 5, the total
ETa values were 138.07 mm, 126.07 mm, 59.54 mm, 145.27 mm and 31.42 mm and their corresponding daily average
ETa values were 2.30 mm∙day
−1, 4.07 mm∙day
−1, 3.72 mm∙day
−1, 3.93 mm∙day
−1 and 2.62 mm∙day
−1 at the seedling, shooting, heading, filling and maturity stages, respectively. The rainfall was approximately 70 mm, and the amount of irrigation water was approximately 510 mm during the entire maize-growing stage, which was similar to the water loss. In the study area, the
ETa was primarily derived from irrigation and was greatly influenced by irrigation events. The cropland was irrigated with approximately 150 mm on 3 June, 120 mm on 25 June, 120 mm on 28 July, 120 mm on 25 August, and 150 mm on 1 November, and the total irrigation water was approximately 660 mm in 2008 [
54]. After the four intervals of irrigation in the maize-growing stage, the soil moisture content (
smc) at a depth of 10 cm depth exhibited a peak (the
smc data on 25 August when the fourth irrigation in the maize-growing stage were lost), and
ETa also increased. After irrigation on November 1, after the maize harvest, the
ETa decreased slightly due to the small
Rn, and because of the lower temperatures, the water in the soil was frozen in winter and stored for the next spring sowing.
During the observation period, the ETa increased in spring, reached a maximum in summer, decreased in autumn and then reached a minimum in winter. This phenomenon may occur because the Rn was low in winter, and the Ta was negative when the soil was frozen. When these conditions occurred, the soil moisture content decreased to the lowest values. The water permeability of the frozen soil layer weakened, which led to a weak relationship between the frozen soil layer and thawed soil layer, and produced low ETa values. Thus, the specific humidity reached a minimum. In spring, however, the Rn was higher and the Ta was positive when the soil thawed. Under these conditions, the soil moisture content increased, which led to increased ETa. In summer, the growth of vegetation flourished, the Rn reached a maximum, the wind velocity was higher, and the cornfield was irrigated four times. Under these conditions, the ETa and the specific humidity reached a maximum. Although the overall rainfall amount was low, the rainfall amount was greater in summer, thereby contributing to the maximum ETa in summer. In autumn, the Rn and Ta decreased, and the surface was bare without vegetation; thus, the ETa began to decline.
During the observation period, the total rainfall was 117 mm. The ETa values were considerably higher than the rainfall, thus leading to arid conditions. The ETa was greatly influenced by the irrigation events and meteorological elements. When the site was irrigated, the ETa peaked the following day.
A regression analysis indicated that the
ETa is closely related to the net radiation, wind velocity, air temperature and specific humidity (
Figure 7) as follows:
where
Rn is the net radiation (MJ∙m
−2∙d
−1),
WS2m is the wind velocity 2 m above the ground surface (m∙s
−1),
Ta2m is the air temperature 2 m above the ground surface (°C), and
q2m is the specific humidity 2 m above the ground surface (g∙kg
−1). The multiplex correlation coefficient was approximately 0.91, whereas the number of cases was 313. Evapotranspiration was positively correlated with net radiation, wind velocity, air temperature and specific humidity. The regression formula (17) showed that wind velocity and net radiation play a significant role in evapotranspiration. When the relationships of evapotranspiration with meteorological factors were assessed in the upper [
55] and middle [
56] reaches of the Heihe River Basin, the effect of wind velocity was greatest, which is consistent with the results of this paper.
3.3. Seasonal Variations of the Reference Evapotranspiration (ET0)
The seasonal variations of the
ET0 at the Yingke site from November 2007 to January 2009 are presented in
Figure 8. The daily average
ET0 values were 2.84 (spring), 5.27 (summer), 2.09 (autumn) and 0.64 mm∙day
−1 (winter). In 2008, the total
ET0 was 1039.92 mm, and the daily average
ET0 was 2.85 mm. The
ET0 observed in the Zhangye Oasis cornfield was similar to the values observed in a cornfield in the semiarid region of northern India [
57], but was higher than the values observed in the Tanggula region of the Tibetan Plateau, except in winter [
50].
During the observation period, the
ET0 was slightly higher than the
ETa, and the differences were large in summer and autumn, and small in winter and spring. Similar to the
ETa, the
ET0 increased in spring, reached a maximum in summer, decreased in autumn and reached a minimum in winter. During the entire maize-growing stage (from 20 April to 22 September 2008), the total
ET0 was approximately 706 mm with a daily average
ET0 of 4.53 mm∙day
−1. As shown in
Table 5, the total
ET0 values were 219.56 mm, 167.97 mm, 76.55 mm, 190.08 mm and 51.76 mm, and their corresponding daily average
ET0 values were 3.66 mm∙day
−1, 5.42 mm∙day
−1, 4.78 mm∙day
−1, 5.14 mm∙day
−1 and 4.31 mm∙day
−1 at the seedling, shooting, heading, filling and maturity stages, respectively. The
ET0 was primarily impacted by meteorological elements and was not influenced by irrigation.
3.4. Crop Coefficient (Kc)
The crop coefficient
Kc was estimated according to FAO56 [
27]:
In 2008, the
Kc ranged from 0.31 to 0.81 (
Table 6), with the maximum values occurring in July, and the minimum values occurring in January. The annual average was 0.56, and the seasonal averages were 0.53 (spring), 0.74 (summer), 0.57 (autumn) and 0.38 (winter).
The Kc values were less than 0.5 outside of the maize-growing stage, because the cornfield was bare without vegetation and not irrigated, except for several rainfall events in April and one irrigation in November.
The
Kc values were greater than 0.5 during the entire maize-growing stage (from 20 April to 22 September 2008) because the growth of corn flourished, and the cornfield was irrigated four times. As shown in
Table 5, the
Kc values were 0.63, 0.75, 0.78, 0.76, 0.61 and 0.71 at the seedling stage, shooting stage, heading stage, filling stage, maturity stage and the entire growth stage, respectively. Li et al. [
25] reported that maize
Kc values in Wuwei City, Gansu Province of northwestern China, at the seedling, shooting, heading, filling, and maturity stages were 0.44, 0.95, 1.46, 1.39, and 1.22, respectively, which generally were higher than our results, except at the seedling stage. The differences are mainly caused by two factors. (1)
Kc is related to vegetation coverage [
27]. In the study by Li et al. [
25], maize was sown with 40 cm row spacing and 6.7 cm planting spacing, and thus the planting density was higher, consisting of approximately 374,800 plants ha
−1. This higher density led to a higher
Kc in the middle and late periods of the growing season. By contrast, in the present study, the maize was sown with a 60 cm row spacing and 25 cm planting spacing, resulting in a lower planting density of approximately 67,000 plants ha
−1 and, consequently, a lower
Kc in the middle and late periods of the growing season. (2) The
ET0 estimated by the FAO56 model was underestimated in the middle and late maize-growing seasons in the study by Li et al. Kang et al. [
58] observed a similar underestimation of
ET0 based on FAO56 in the Loess Plateau, Shaanxi, China, and a higher
Kc compared with the values given by Allen et al. [
27]. This phenomenon also led to a higher
Kc in the middle and late growing seasons.
In recent years, other methods of studying the
Kc of maize have been used, such as models and remote sensing methods. Miao et al. [
59] focused on the actual evapotranspiration, crop transpiration and crop coefficient using the SIMDualKc model which is a model for simulating soil water balance based on FAO56 double crop coefficient method in the Hetao irrigation district of the upper Yellow River basin, China, and a new modelling approach was developed for the basal crop coefficients (
Kcb) of a relay-strip intercropping system. Kullberg et al. [
60] compared several remote sensing methods to calculate crop evapotranspiration and
Kcb in a deficit irrigation experiment for maize near Greeley, Colorado, and the results showed that remote sensing methods can inform users about the availability of certain data and irrigation levels. Models and remote sensing methods are important methods in regional evapotranspiration and
Kc research [
61,
62,
63].