Estimation of irrigation requirements for drip-irrigated maize in a sub-humid climate

LIU Yang, YANG Hai-shun, LI Jiu-sheng LI Yan-feng YAN Hai-jun

1 College of Water Resources and Civil Engineering, China Agricultural University, Beijing 100083, P.R.China

2 State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin, China Institute of Water Resources and Hydropower Research, Beijing 100048, P.R.China

3 Department of Agronomy and Horticulture, University of Nebraska-Lincoln, Lincoln 68503, United States

1.Introduction

Heilongjiang Province has the largest maize area and production in China, accounting for 15 and 16% of national maize area and production, respectively (NBSC 2015),playing an important role in national food security.In Heilongjiang Province, the dominant climate is temperate sub-humid continental monsoon, where winter is long, cold,and dry with a short but warm and wet summer growing season.The total seasonal precipitation in Heilongjiang Province usually meets the water demand for maize in most years but poor rainfall distribution in relation to crop water demand often leads to crop water stress at critical stages (e.g., kernel setting, grain filling, etc.), resulting in reduced yields.Less than 10% of the maize sown area is irrigated and on-farm maize yields were, on average,only 51% of the potential yields in this region (Liu Z J et al.2012, 2016).Moreover, the rain-fed maize yield is low and unstable in areas with lower precipitation (Liu et al.2016).Consequently, effective irrigation could improve maize production and narrow yield gaps between rainfed and irrigated conditions in Northeast China (Liu Z J et al.2012;Liu C et al.2017).

Drip irrigation is one of the most efficient methods of irrigation/fertigation in terms of application efficiency and reducing soil evaporative losses (Irmak et al.2016).In recent years, drip irrigation has widely been applied to maize production in sub-humid regions like North China Plain (Wang et al.2014), Northeast China (Liu et al.2015),and Central U.S.(Lamm and Trooien 2003; Irmak et al.2016) due to its advantages of precise application in amount and at location throughout the field and effectiveness in improving water and nitrogen use efficiency compared to other irrigation methods (Bar-Yosef 1999; Guan et al.2013).After ten years of research in Kansas in the U.S.,Lamm and Trooien (2003) concluded that irrigation water used for corn can be reduced by 35 to 55% when using subsurface drip irrigation compared with traditional irrigation.For drip-irrigation management in the field (e.g., irrigation frequency, amounts), several methods are commonly used including readings from soil moisture sensors (Leib et al.2003), monitoring of crop water stress index (Jackson et al.1981), and estimating crop evapotranspiration (Allen et al.1989).Although those methods can be used at field level, they do not allow easy estimation of regional irrigation requirements at larger spatial scales, e.g., for a province like Heilongjiang Province due to variations in climate, crop systems, management practices and soil types.

Crop growth modeling can potentially be a good method to estimate the water and nutrient managements under varying weather and soil conditions (Boote et al.1996).Some simulation models (e.g., CERES-Maize,AquaCrop, APSIM, RZWQM, Hybrid-Maize) have been tested to simulate crop yield, evapotranspiration and water management strategies for maize in arid or semi-arid regions(Abedinpour et al.2012; Jiang et al.2016).Abedinpour et al.(2012) evaluated the performace of the FAO AquaCrop model for maize crop in a semi-arid region and the results showed that the model predicted maize yield with acceptable accuracy under variable irrigation and nitrogen levels.The Hybrid-Maize model (Yang et al.2014, 2016) has also been widely tested under rainfed and irrigated conditions and applied to the U.S.corn-belt (Grassini et al.2009, 2011;Morell et al.2016), South Asia (Timsina et al.2010), and North China (Hou et al.2014a; Bu et al.2015).Liu Y et al.(2012) evaluated the Hybrid-Maize model to simulate maize growth and yield in a semi-arid Loess Plateau and applied the model to assess effects of meterological variations on the performance of maize under rainfed and irrigated conditions.According to the simulations, the average rainfed yield was 1 830 kg ha-1less than the average potential yield with irrigation.In contrast, there were few studies that have used models to simulate water and nitrogen strategies for maize in sub-humid regions (Liu et al.2013; Zhang et al.2015).Jiang et al.(2016) used long-term weather data to simulate the effects of different irrigation treatments on maize yield and water use efficiency and recommended the total irrigation amounts regardless of the rainfall each season.Using the calibrated CERES-Maize model, He et al.(2012) identified the best irrigation management practices for sweet corn production on sandy soils, which indicated that irrigation frequency had a strong influence on sweet corn yield.However, crop water requirements varied from different physiological stages and the effects of water stress on growth and yield during different growth stages might also differ (Jones and Kiniry 1986; Kozak et al.2005).Liu Y et al.(2017) simulated the sensitivity of maize to water at varied stages and the simulation results indicated that the descending order was pollen shedding and silking,tasselling, jointing, initial grain filling, germination, middle grain filling, late grain filling, and end of grain filling.In Florida, He et al.(2012) found corn growth suffered water stress and the simulated yield was reduced if irrigation events were triggered when the maximum allowable depletion of soil water content was greater than 60%.In practice, a substantial number of fields (55% of total) had water supply in excess of that required to achieve yield potential (Grassini et al.2011).Analysis results in the Western U.S.Corn Belt also indicated that up to 32% of the annual water volume allocated to irrigated maize in the region could be saved with little yield penalty (Grassini et al.2011).Such research on estimating irrigation requirements during mazie water-sensitive stages was helpful to reduce water supply and improve irrigation schedules to be more synchronous with crop water requirements.

For regional upscaling, irrigation requirements (e.g.,irrigation timing and amounts) could be estimated with consideration of soil water content at sowing stage, crop water requirements at different stages, crop management practices, cultivar maturity, plant population, soil type, and climate characteristics at diverse agro-climatic zones for providing irrigation guidance (Amarasingha et al.2015;He and Cai 2016).However, data collection at a large number of locations is expensive and time-consuming.The minimum number of locations was required to achieve robust estimates at larger spatial scales.An issue is the ability of crop models to predict local and regional actual yield and total production without need of site-year specific calibration of internal parameters associated with fundamental physiological processes (Morell et al.2016).van Bussel et al.(2015) described an approach that consists of a climate zonation scheme supplemented by agronomical and locally relevant weather, soil and cropping system data.Variation in simulated yield potentials among weather stations located within the same climate zone can be represented by the coefficient of variation and served as a measure of the performance of the climate zonation scheme for upscaling(van Bussel et al.2015; Morell et al.2016).Therefore, crop simulation models can be used to predict local to regional maize yields and total production (Morell et al.2016).In the same way, more research on scaling up location-specific drip-irrigation requirements estimates under diverse agroclimatic zones will assist establishment of better dripirrigation management strategies for maximizing maize production in China.

The objectives of this study were to: (1) estimate drip-irrigation requirements during different physiological development stages of maize using model simulation, and(2) evaluate the difference of irrigation amounts for dripirrigated maize under diverse agro-climatic conditions in sub-humid region of Northeast China.

2.Materials and methods

2.1.Field experiment

A field experiment was conducted for three years (2011,2012 and 2013) at a research experimental station (45°22´N,125°45´E, 220 m above sea level) located in Harbin,Heilongjiang Province, Northeast China.The region has a sub-humid climate with a long-term (from 1980 to 2010)average seasonal (May to September) mean air temperature of 20.5°C and average seasonal rainfall of 421 mm.The dominant soil texture is silt (Shirazi and Boersma 1984)(Table 1).At three locations of the field, undisturbed soil samples were taken at three depth intervals (0 to 20 cm, 20 to 40 cm, and 40 to 80 cm) for measurements of bulk density,field capacity following the method by Veihmeyer and Hendrickson (1949), and wilting point at 1.5 MPa pressure using a centrifugal method (CR 21GII, Hitachi, Japan)(Table 1).Daily weather data, including the maximum and minimum temperatures, relative humidity, wind speed, and sunshine hours were obtained from an automatic weather station located approximately 500 m from the experimental field while rainfall data were collected manually from four rain gauges installed at each corner of the field.

Prior to planting, the field was prepared to have ridges of 1 m wide with 0.3 m wide furrows in between (Fig.1).Two rows of maize were seeded on each ridge with a spacing of 0.5 m.Each plot had eight rows of maize.Maize was planted on May 5 in 2011, May 4 in 2012, and May 9 in 2013.A similar plant spacing of 0.33 m along a row was used for the three growing seasons, and the resultant plant density was about 46 620 plants ha-1.After planting and before emergence, a dripline was laid in the middle of two rows on each ridge and a 1.2 m-wide strip of plastic film of 0.008 mm thick was laid to cover the driplines and the soil surface(Fig.1).Immediately after emergence, an opening of about 5 cm in diameter was manually punched in the plastic film at the position where a plant emerged to allow the plant to come through the mulch.Pest and weed control followed conventional practices in the region.The maize was harvested on September 15 in 2011, September 27 in 2012,and September 25 in 2013.After harvest, plastic films and maize stalks were removed from the field (Liu et al.2015).

The emitters of the drip lines had a spacing of 0.3 m(IrriGreen Ltd., Beijing, China) and a nominal flow rate of 2.0 L h-1at 0.1 MPa.For irrigation management, a target wetting depth of 40, 50, 70 and 60 cm was used for the initial(emergence to 6-leaf), establishment (6-leaf to tasseling),mid-season (tasseling), and late season stages (effective grain filling), respectively (Allen et al.1998).Irrigation was applied whenever average soil water content in the target wetted depth depleted to around 60% of the field capacity(Liu et al.2015).The amount of irrigation was determined to replenish to 85% of the field capacity of the target wetting soil depth.The field received 349 mm of rainfall and 35 mm of irrigation in 2011 growing season, 515 mm of rainfall and 70 mm of irrigation in 2012 growing season, and 569 mm of rainfall and 45 mm of irrigation in 2013 growing season.Compared with 30-year (1981 to 2010) historical seasonalrainfall of the same period, it was wet for each season except 2011 season (Liu et al.2015).All plots received a basal application of 54 kg ha-1of N and 138 kg ha-1of P2O5in the form of diammonium phosphate and 81 kg ha-1of K2O in the form of potassium sulfate prior to planting in the 2011 and 2012 seasons, but no basal fertilizers in 2013.Besides the basal application, a total of 150 kg ha-1of N of urea was applied through drip irrigation equally during the 8- to 12-leaf stage, tasseling, and blister (R2) stages during each season (Liu et al.2015).

Table 1 Basic soil properties of experimental field

Fig.1 Schematic diagram of the cropping pattern and lateral layout of the driplines under the plastic mulch used for maize.

For each season, soil samples were taken at five depths of 0 to 10 cm, 10 to 20 cm, 20 to 40 cm, 40 to 60 cm, and 60 to 80 cm in each plot 12 days after planting as well as at harvest to obtain the initial and final soil water contents,respectively.Specifically, the soil samples were taken from the middle of two central rows of each plot.Soil samples at depths of 0 to 10 cm, 10 to 20 cm, 20 to 40 cm, 40 to 60 cm, and 60 to 80 cm were also collected three to seven days before and after each irrigation event to obtain the seasonal change of water content in the soil.Soil samples were dried at 105°C to a constant weight to determine gravimetrical water content.In this study, soil water content of the total profile (0 to 120 cm) was calculated by accumulating soil water content of each layer.The average soil water content at depth of 80 to 120 cm was assumed to be the same as the average at depth of 60 to 80 cm due to minor difference beyond 60-cm depth based on experimental observations (Liu Y et al.2017).

Plant height and leaf area index (LAI) were measured in three 13-m sections of the four center rows in each plot.In each section, three average plants were marked for the measurement of plant height and LAI at jointing, silking and around blister stages.For LAI measurements, the length and the maximum width of each leaf were recorded.In addition, the actual area for 15 typical leaves selected other than the marked plants were measured using coordinate grids.A linear regression between the actual area and the product of the length and width of the leaf was obtained for each measurement.The product of the leaf length and width for the three marked plants was then converted to the actual leaf area using the linear regression model.Finally,LAI was calculated by dividing the total actual leaf area of the three marked plants by the ground area.

For aboveground biomass, three average plants were collected in each plot by clipping the plant at the soil surface.The stalks and ears of three plants were harvested separately in each plot at maturity.All plant samples were oven-dried at 70°C to a constant weight (Liu Y et al.2017).

For grain yield (GY) determination, maize ears were hand-harvested from four approximately equally distributed locations of six consecutive plants per location (totally 24 plants) in each plot and grain yield was expressed at a moisture content of 14%.

2.2.Model description

The Hybrid-Maize model is a process-based model that simulates maize development and growth on a daily time-step under growth conditions without limitations from nutrient deficiencies, toxicities, insect pests, diseases, or weeds (Yang et al.2004, 2006).The Hybrid-Maize model requires daily weather variables including solar radiation,the maximum and minimum air temperatures to simulate corn stages and dry matter accumulation and requires precipitation, wind speed and humidity in order to simulate crop water uptake and soil water balance.In Hybrid-Maize model, photosynthetically active radiation interception(PARi) and gross assimilation are described according to formulations in WOFOST (Boogaard et al.2014).The PARiand its corresponding CO2assimilation are computed for each layer in the canopy.Total gross assimilation is then obtained by integration over all layers.Using L to represent the depth of canopy with L=0 at the top and L=LAI at the bottom of the canopy, the PARiat position L in the canopy equals the decrease of PAR at that depth.Calculation of PAR was as eq.(1):

Where, PARi,Lis the PAR interception by the canopy layer at position L, I is the incoming total solar radiation and k is the light extinction coefficient.The corresponding CO2assimilation by that layer follows a saturation function of the form:

Where, ALis the CO2assimilation by the canopy layer at L, Amis the maximum gross CO2assimilation rate (g CH2O m-2leaf h-1), and ε is the initial light use efficiency (g CO2MJ-1PAR).The CO2assimilation by the whole canopy is obtained by integration of eq.(2) along L:

Where, A is the gross CO2assimilation of the canopy(g CO2m-2ground h-1).Two numerical integration methods are available in the model.The default method, which was used in all the simulations of this study is the three-point Gaussian method (Goudriaan 1986).Alternatively, a user can choose the standard Simpson’s rule with a user-defined precision.

In the special version of the Hybrid-Maize model for this study, the heating effect by plastic film mulching was taken into account for growing degree days (GDD) above 10°C(GDD10) accumulation before 6-leaf stage as the maize growing point remains below soil surface until then (Ritchie et al.1992; Hou et al.2014a; Liu Y et al.2017).After 6-leaf stage, no heating effect of plastic film mulching is considered as the maize growing point has risen above soil surface and is supposed to be outside the plastic film.

The model simulates separately soil evaporation and crop transpiration, as well as other losses including surface runoff, canopy interception, and drainage below crop rooting depth.The model simulates progression of crop rooting depth based on GDD accumulation and the maximum rooting depth is reached shortly after silking.The crop is assumed to take up water only from the active rooting zone and crop water uptake is related to water content and hydraulic conductivity.The whole rooting zone is divided into layers of 10 cm, and water balance is computed layer by layer from the top to bottom (0 to 120 cm) based on the principle of tipping bucket method (Yang et al.2004).

Actual soil evaporation is estimated using the 2-step evaporation scheme as Allen et al.(1998).According to this scheme, soil evaporation occurs within the top 10 cm soil depth, and the evaporation rate will be constant at its maximum when soil is wet (i.e., more than 70% of readily evaporate water), followed by a decreasing rate before evaporation ceases at the half of permanent wilting point.Considering the plastic film breakage (including punching holes for emergence) during the growing season, average soil surface coverage rate of the plastic film mulching treatment is set as 50% of bare soil (Liu et al.2015).Crop actual transpiration (Transpactual) is the smaller one between the maximum water uptake by roots from all layers where roots are present and the maximum demand for transpiration(Transpmax) estimated from weather conditions (Yang et al.2004).

For simulating irrigation requirements with drip irrigation,irrigation was called in the model whenever crop water stress starts to appear on a daily basis.Water stress index was expressed as:

The crop suffers no stress and full stress when water stress equals 0 and 1, respectively.

The maximum amount of water that can be applied in each irrigation event was set at 30 mm for drip irrigation in this study and the irrigation target soil water content in top 30 cm was set at 85% of the field capacity.

2.3.Model calibration

The Hybrid-Maize model was calibrated using the observed data of 2012 for soil water content over the rooting depth,LAI, aboveground dry matter, and grain yield.The potential kernel number per ear and light extinction coefficient were selected for calibration, because they are more hybridspecific and the model’s default values are more suited to North American hybrids instead of those common in China.Maize hybrids in North American are more suited to higher maize plant densities (more than 60 000 plants ha-1) and tend to have smaller and more vertical leaves, and smaller ears and fewer kernels, while hybrids of smallholder fields in China are more suited to lower densities (less than 60 000 plants ha-1) and tend to be the opposite in terms of leaf angle and ear size (Russel et al.1989; Girardin and Tollennaar 1994; Otegui 1995; Shi et al.2016).

For the potential number of kernels per ear, the default value of 675 was increased to 800 for better simulation results for hybrids used in this study (Yang et al.2004).Such an adjustment was also suggested by Jones and Kiniry(1986).Similarly, the default light extinction coefficient (k)of 0.55 was calibrated to 0.75, which is still within the range of possible value for maize k (Maddonni et al.2001; Lizaso et al.2003; Lindquist et al.2005).The calibrated model was then tested and validated using data of 2011 and 2013.

2.4.Model application

Estimating irrigation requirements during different growth stages at the experimental siteThe calibrated Hybrid-Maize model was applied to estimate the irrigation requirements (irrigation dates and amounts) during different crop stages for mulched and drip-irrigated maize using 30-year historical weather data (1981 to 2010) at the experimental site.The historical weather data were acquired from local meteorological bureau whose station was within 20 km from field.The simulated results included daily maize growth variables, crop stages, LAI, total biomass, crop evapotranspiration (ETc), irrigation requirements (irrigation dates and amounts), and final grain yield.The initial soil water available content (ISWC) was set as 20, 40, 60, 80,and 100% of the maximum soil available water content in the root zone (the soil water content between field capacity and permanent wilting point), respectively.The date of the first drip-irrigation event was also simulated and analyzed.The hybrid-specific input parameters of the applied model were the same as the field experiment in this study.The planting time was set as May 1 for all 30-year simulations because local farmers usually planted maize around the period from the end of April to the start of May.

Besides the phenology stages based on leaf numbers and kernel filling progression, crop development stage was also expressed using a dimensionless scale from 0 (planting time) to 1.0 (tasseling) to 2.0 (physiological maturity)(Lindquist et al.2005).Before to silking, the numerical scale stage is the ratio of up-to-date total GDD since planting to the total GDD at silking; after silking, it is the ratio of the upto-date total GDD since silking to the total GDD from silking to physiological maturity plus one.The correspondence of the phenology stages and the numerical stage is: planting to 6-leaf stage (V6; 0 to 0.43), V6 to 10-leaf stage (V10;0.43 to 0.71), V10 to tasseling stage (VT; 0.71 to 1.00), VT to milk stage (R3; 1.00 to 1.27), R3 to dent stage (R5; 1.27 to 1.67), and R5 to physiological maturity (R6;1.67 to 2.00).In this study, the kernel setting window corresponds to the numerical stage of 0.87 to 1.13.The kernel setting window(about 4-week bracketing silking), one of the most watercritical stage for maize, was considered specially in this study.The number of kernels was determined during this period, influencing the potential size of storage organ (i.e.,the sink).The maize can lose kernels permanently due to water stress during this stage.In the Hybrid-Maize model,the kernel setting window was defined from 170 GDD8(i.e.,8°C based) before silking to 170 GDD8after silking (Yang et al.2004, 2006).

The effects of water stress during different crop stages on grain yield and aboveground biomass were also studied through five drip-irrigation scenarios: (1) full irrigation, (2)no irrigation before kernel setting window, (3) no irrigation during kernel setting window, (4) no irrigation after kernel setting window, and (5) no irrigation at all (i.e., rainfed condition).For simulating the effects on different rainfall distributions, six typical weather years in the experimental area were chosen from 30-year historical weather data according to seasonal rainfall amounts, including two dry years (1989 and 2007), two normal years (1997 and 2003),and two wet years (1987 and 1998).The ISWC was set as 40% in this part of study.

Classification of agro-climatic zones in Heilongjiang ProvinceGeospatial distributions of harvested areas of maize in Heilongjiang Province (Fig.2-A and B) were derived from the global Spatial Production Allocation Model(SPAM2005, You et al.2014).SPAM2005 provides gridded data (five arcmin resolution, approximately 10 km×10 km at the equator) on annual harvested area averaged for years around 2000 for 20 major staple crops.SPAM2005 was selected because it applies a consistent methodology using available data on harvested crop area to derive global spatially disaggregated harvested area maps (van Bussel et al.2015).

Fig.2 Schematic diagram of the harvested area of maize (A) and agro-climatic zones distribution (B) in Heilongjiang Province,China.T and W indicated levels of growing degree days (GDD) and arid index, respectively.The GDD and the arid index became greater when T increased from T1 to T3 and W decreased from W5 to W1, respectively.

In order to up-scale location-specific estimates of maize yield and seasonal irrigation requirements to regional levels,the major maize areas were divided into agro-climatic zones(CZs) according to the method by van Wart et al.(2013).A matrix of three categorical variables were used to delineate CZs for harvested area of maize: GDD, arid index (W), and temperature seasonality.Consequently, main maize harvest areas in Heilongjiang Province was divided into 10 CZs(Fig.2).Among them, there were three levels of GDD (T1 to T3), seven levels of W (W1 to W7) and one temperature seasonality.The GDD and the W became greater when T increased from T1 to T3 and W decreased from W5 to W1,respectively.In each CZ, the maize production (MP, kg)was estimated by:

MP=GY×HA (5)

Where, GY was the grain yield (kg ha-1) and HA was the harvested area (ha).

In this study, 24 weather stations were selected according to the method described by van Bessel et al.(2015).The climate data were obtained from the National Meteorological Networks of China Meteorological Administration (http://cdc.cma.gov.cn) (Fig.2).Each weather station was identified when the sum of maize harvested area within a 100-km radius of each weather station in this CZ were above 50%of the total maize harvested area of this CZ (van Wart et al.2013; Grassini et al.2015).Daily weather variables were acquired from 1981 to 2010 including the daily maximum and minimum air temperature, relative air humidity, precipitation,sunshine duration, and average wind speed.Sunshine duration was converted into daily solar radiation using the Ångström formula method (Jones 1992).

Hybrid-maize information surrounding each weather station was acquired according to Hou et al.(2014b).Four GDD maturity levels from total GDD of 1 150 to 1 580°C days were used in different CZs of Heilongjiang Province(Table 2).GDD was defined by:

Where, n is days from planting to maturity, Tmax, Tmin, and Tbaseare the maximum temperature, minimum temperature,and 10°C base temperature, respectively (McMaster and Wilhelm 1997); a upper cut-off of 30°C is used to set Tmaxif it is greater than 30°C.

The planting date was uniformly set as the same day for all 30-year simulations on one site but differ across sites.On each site, the planting date is when the average air temperature of above 10°C last for one week in the late April or early May to guarantee the emergence of the seeds.Maize growth will terminate when it comes to maturity, or by frost or severe water stress.

The soil data were extracted from the National Soil Atlas of China (1:14 000 000, ISS 1986).The soil data contained bulk density of the topsoil, and texture of top and subsoil,texture, pH and soil organic carbon (SOC) content (i.e.,30 cm in depth) (Table 2).Within the 100 km-diameter scope of each weather station, the dominant soil type was selected to represent for an area (Table 2).

2.5.Statistics analysis

Three statistics indices were used to evaluate the simulation results against field measurements: (i) root mean squared error (RMSE) as defined in eq.(7); (ii) relative RMSE(RRMSE, %) as defined in eq.(8); and (iii) index of agreement (d-index) as defined in eq.(9), which ranges from 0 to 1 with 1 representing a perfect fit:

Where, Oiand Piare the observed and predicted values,Oavgis observed averages and n is the number of values.

3.Results

3.1.Model performance

The calibrated Hybrid-Maize model performed reasonably well for simulating total soil water content in the root zone,LAI, and aboveground dry matter accumulation in the three growing seasons from 2011 to 2013.The RRMSE and d-value were less than 25% and above 0.9, respectively which were both in the range of acceptance (Table 3).But the model overestimated the LAI during the early growing season and underestimated the aboveground dry matter at maturity (Fig.3).The reason of overestimation of the LAI might be the function of leaf area expansion in the Hybrid-Maize model may not fully reflect the cultivars used in China.The reason for the underestimation of dry matter at maturity might be that the Hybrid-Maize model was developed and calibrated (other than the parameters calibrated in this study)largely for high plant density systems in North America,leading to underestimation of aboveground dry matter at maturity for lower density systems in Northeast China.For grain yield, the calibrated model did well for 2011 and 2012 growing seasons but overestimated for 2013 (Fig.3)with the d-value being only 0.38 (Table 3), which might be because the Hybrid-Maize model simulates maize growth under optimal management conditions and as a result it often overestimates crop growth, including leaf area index and biomass.However, in the field experiments, there might be nutrition deficiencies, especially for nitrogen as introduced by nitrate leaching.As mentioned earlier, no basal fertilizer was applied at planting in 2013 and the initial soil water content at planting was pretty high due to melting snow of the last winter, which might lead to nitrate leaching and deficiency (Liu et al.2015).

3.2.Estimating irrigation requirements at the experimental site

Growing season precipitation and ETcThe growing season precipitation varied from 302 to 786 mm and were lower than the ETcunder fully irrigation conditions that varied from 467 to 727 mm in 29 out of the 30-year simulation at the experimental site (Fig.4).The greatest difference between the growing season precipitation and ETcwas 323 mm occurring in 1999 and the difference was above 200 mm in nine years out of the 30-year simulation(Fig.4).The average growing season ETc(607 mm) was 32% (146 mm) greater than the average growing season precipitation (461 mm).It implied that the average 146 mm of water requirement for a water stress free maize crop should come from either soil moisture storage present at planting or supplemental irrigation.As a consequence, in other words, the irrigation requirements depend highly onthe amount of initial soil moisture status at planting.

Table 2 Hybrid-maize information and soil properties in agro-climatic zones of Heilongjiang Province, China

Table 3 Statistic analysis of Hybrid-Maize model performance on simulating mulched and drip-irrigated maize of 3-year data1)

Fig.4 Total precipitation and total crop evapotranspiration(ETc) of fully irrigated maize in 30-year (1981 to 2010) growing seasons at the experimental site.

Irrigation requirements during different crop stagesThe simulated on-avearage seasonal irrigation amounts for mulched and drip-irrigated maize decreased from 150 to 48 mm with ISWC increasing from 20 to 100% of the maximum soil available water capacity at the experimental site (Table 4).When ISWC was lower than 40%, the maize might need one drip-irrigation of 10 to 30 mm before V6 stage due to occasional spring drought during the seedling establishment while no irrigation is needed when ISWC was greater than 40% (Table 4).

During V6 to V10 stages, there was much possibility to irrigate regardless of the level of ISWC.When ISWC was lower than 40%, there were 34 to 41 mm of irrigation requirements on average (Table 4), while on average 6 to 22 mm of water is required during this period when ISWC was greater than 40%.From V10 to R3 stage, there was on average 14 to 36 mm of irrigation water when ISWC varied from 20 to 100% (Table 4).From R3 to R6 stage, there was on average 28 to 48 mm of irrigation water if ISWC increased from 20 to 100% (Table 4).For kernel setting window, there was on average 6 to 15 mm of irrigation water.Among the 30 years from 1981 to 2010, the maximum requirement of irrigation amounts during kernel setting window was 1999 with 62 mm of water (data not shown).

First drip-irrigation eventThe first drip-irrigation event varied from May 5 to June 20 in the 30-year simulations with the ISWC differing from 20 to 100%.The dates of the first drip-irrigation event moved later into the season when the ISWC increased from 20 to 100% (Table 5).The average date of the first drip-irrigation event varied from June 7 to July 27 with the ISWC increasing from 20 to 100%.In terms of crop stage, the first drip-irrigation event varied from V3 to V8 with the ISWC increasing from 20 to 100%, while the average crop stage of the first drip-irrigation event moved backward from V5 to silking with the ISWC increasing from 20 to 100%.

The effects of water stress at different stages on grain yield and aboveground biomassThe grain yield and final aboveground biomass under rainfed conditions decreased due to water stress (Fig.5).On average, 73, 52, and 30% of grain dry matter was lost under rainfed conditions compared to using drip-irrigation in dry, normal, and wet years, respectively (Fig.6).In dry year (1989), the crop can lose more grain yield due to prolonged water stress at critical stages as crops under rainfed conditions stopped growth and became pre-matured (Fig.6).Even in normal year like 1997, the crops can lose significant grain yield due to severe water stress during kernel setting window and resulting in decreased kernel numbers.

Grain yield was affected if no irrigation was provided before kernel setting window (vegetative stage), especially when the rainfall was less well distributed during vegetative stages like 1997 (Figs.5 and 6).However, the effects of no irrigation during vegetative stage were greater on aboveground biomass than grain yield.For example, in 1997, water stress without irrigation before kernel setting window led to a loss of 33% (7.0 t ha-1) in total biomass but 23% (3.1 Mg ha-1) in grain yield (Fig.6).

Irrigation during kernel setting window was critical when the ISWC was relatively low and rainfall during vegetative stage was less (Fig.5).For example, even in the wet year of 1998, the crops lost 13% (1.6 t ha-1) of grain yield when no irrigation was given during kernel setting window, resulting in a drought during this period (Fig.6).

Irrigation was very critical to grain filling (reproductive stages), especially when there were little rainfall distributions during this period (Fig.6).For example in the dry year of 2007, without irrigation after kernel setting window led to a yield loss of 53% (5.6 t ha-1) due to water stress during this period (Fig.6-A).

3.3.Evaluating the irrigation requirements for drip-irrigated maize in diverse climatic conditions in Heilongjiang Province

Growing season characteristics of different agro-climatic zonesThe growing season period varied from 135 to 164 days across different CZs of Heilongjiang Province (Table 6).With GDD10increasing from T1 to T3, the growing period became longer from 138 to 161 days.The CZs of T3W1 and T3W2 had the longest growing days (161 to 164 days)and greater GDD10(1 580°C days) due to warmer climate.In contrast, the CZs of T1W3, T2W5 and T1W5 had the shortest growing days (135 to 136 days) and the lowestGDD10(1 150 to 1 310°C days) because of relatively cooler climate.

Table 4 Thirty-year average irrigation requirements (means±SD) during different crop stages for the experimental site estimated by the Hybrid-Maize model

Table 5 First irrigation event and corresponding development stages (DVS) with initial soil water content (ISWC) varying from 20 to 100% using 30-year historical weather data (1981 to 2010) in the experimental area

Fig.5 The dynamic process of water stress with crop development stage under four scenarios.Crop development stage was expressed using a dimensionless scale from 0(planting time) to 1.0 (tasseling) to 2.0 (physiological maturity).KSW, kernel setting window.A, dry year (1989).B, normal year (1997).C, wet year (1987).

From W1 to W5, the seasonal precipitation increased from 361 to 496 mm and ETcdecreased from 645 to 405 mm(Table 6).All CZs except T1W5, T2W5, and T1W4 had less seasonal precipitation than ETc(Table 6).This implies that most of the maize area in Heilongjiang Province requires irrigation.Among all the CZs, T3W1 had the lowest precipitation (361 mm) and the largest ETc(645 mm), which means that at least 285 mm of water must be provided either from soil water in the root zone or irrigation in order to achieve the maximum grain yield.On the contrary, the T1W5 zone had the largest average seasonal precipitation(501 mm) and the lowest ETc(363 mm) due to its location in a mountainous area.

Irrigation requirements in different agro-climatic zonesTen CZs were divided into three levels according to the degree of irrigation requirement (Table 7).In the CZs of T3W1,T3W2 and T2W2, which need a large amount of irrigation,at least 80 mm of irrigation was needed regardless of the ISWC (Table 7).In the CZs of T3W3, T2W3, and T2W4 which need moderate rates of irrigation, at least 30 mm of irrigation was needed to achieve the highest grain yield even with the 100% of ISWC.For the CZs of T2W5, T1W3, T1W4 and T1W5, little or no irrigation was needed at optimal ISWC.

2）以第1节模板作为支撑，进行第2节的钢筋接长、绑扎，验收合格后支立第2节模板，模板加固验收合格后进行混凝土浇筑；

3.4.Effects of irrigation on maize production in different agro-climatic zones

The effects of irrigation on maize production were not only related to grain yield but also to maize production area.The CZs of T3W4, T2W3 and T2W4, which require moderate rates of irrigation had the largest harvested area within all CZs, accounting for 70% of the total harvest area of maize in Heilongjiang Province (Table 8).And the harvested areas of CZs that require large and small amounts of irrigation water accounted for 24 and 6% of the total harvested area,respectively (Table 8).

Among nine out of all ten CZs, the grain yield was greater with irrigated systems compared to rainfed conditions except T1W5 (Fig.7).The effects of irrigation on grain yield were greater with higher demands for irrigation water (Fig.7).For instance, the CZs of T3W1, T3W2 and T2W2 require large amount of irrigation water and their average increase of grain yield using irrigation was 109% (7.1 t ha-1) and 50%(4.6 t ha-1) with 40 and 100% of ISWC, respectively, higher than rainfed yield.In contrast, the CZs of T2W5, T1W3,T1W4 and T1W5 only require a small amount of irrigation and their grain yield was only 10% (0.8 t ha-1) and 2%(0.2 t ha-1) with 40 and 100% of ISWC, respectively, higher than rainfed yield (Fig.7).

For the whole Heilongjiang Province, the total maize production could rise by at least 42 and 14% with irrigation systems with 40 and 100% of ISWC, respectively, compared to rainfed conditions (Table 8).For the CZs of T3W3,T2W3 and T2W4 which require moderate irrigation, yield increase could be 56 and 43% of the total maize-production increase in Heilongjiang Province with 40 and 100% of ISWC, respectively (Table 8).For the CZs of T3W1, T3W2 and T2W2 which require tremendous irrigation, production increase could be 43 and 56% of total maize production in Heilongjiang with 40 and 100% of ISWC, respectively(Table 8).In contrast, only 1% increase of maize production could be increased through irrigation systems in the CZs of T2W5, T1W3, T1W4 and T1W5 which require little irrigation(Table 8).

Fig.6 Grain dry matter (A) and final aboveground biomass (B) under five irrigation scenarios in dry years (1989 and 2007), normal years (1997 and 2003), and wet years (1987 and 1998).KSW, kernel setting window.

4.Discussion

Heilongjiang Province is located at a typical cool highlatitude area (43°26´-53°33´N) with the low mean annual air temperature (from -5 to 5°C), which belongs to tempreate continental monsoon cilmate.Crop productivity depends largely on uneven precipitation in summer and fall (Li and Liu 2006; Song et al.2013).The effects of supplemental irrigation on improving maize yield become more critical due to less precipitation and increased warmer weather(Shi et al.2014).In this study, we found that precipitation in 94% of the maize harvested area did not meet the demand of water for maize in Heilongjiang Province.In addition,without supplemental irrigation at any development stages,water stress will develop and affect grain yield because of poor distribution of precipitation, even in wet years with total precipitation greater than crop evapotranspiration.Moreover, the spatial variation in irrigation requirements is relatively large in sub-humid regions like Heilongjiang Province due to the East Asian summer monsoon and related seasonal rain belts, which had significant variability at intraseasonal, interannual and interdecadal time scales (Li and Liu 2006).Furthermore, the initial soil available water before planting is occasionally low due to less precipitation in winters in monsoon environments.The first supplemental irrigation event usually comes early (sometimes on seedling stages) because of dry winter and little precipitation during the early stage of maize (Table 5).Although the impacts of water stress are greater on grain yield during kernel setting widows and grain filling stages compared to vegetative stages (Fig.5), a significant grain yield reduction can stillresult from drought during vegetative period at early ear shoot and ovule development (Claassen and Shaw 1970).Meanwhile, as a result of global warming, extreme drought in Northeast China is increasingly interfering with the steady development of grain production (Xu et al.2017).Timely irrigation is critical to achieving potential yield in a sub-humid Northeast China.

Table 6 Thirty-year (1981 to 2010) meteorological attributes (means±SD) during the maize growing seasons (May-September)in different agro-climatic zones of Heilongjiang Province, China

Table 7 Irrigation water requirements (means±SD) in 10 agro-climatic zones which were grouped into three levels based on irrigation requirement

5.Conclusion

Crop growth modeling was used in sub-humid environments to estimate the irrigation requirements for drip-irrigated maize during different crop development stages and to evaluate the effects of drip irrigation under diverse agroclimatic conditions in sub-humid region.The following conclusions were supported by this study:

(1) In sub-humid region with summer monsoon, the irrigation requirements during different crop stages were highly related to initial soil water content and seasonal precipitation distributions.A lower initial soil water availability requires a larger amount of irrigation water and an earlier first irrigation event.

(2) The effects of drip irrigation may vary a lot under different climatic conditions.Overall, irrigation was veryimportant for maize production in sub-humid regions like Heilongjiang Province.With drip irrigation, the total maize production in Heilongjiang Province could increase 14 to 42% (3.6 to 8.5 million t) compared to rainfed conditions.

Table 8 Comparison of maize production between irrigated and rainfed conditions in different agro-climatic zones of Heilongjiang Province, China

Fig.7 Comparison of grain yield between irrigated and rainfed conditions with initial soil water content accounting for 40 and 100%of total soil available water in different agro-climatic zones of Heilongjiang Province, China.T and W indicate levels of growing degree days (GDD) and arid index, respectively.Vertical bars are SD between historical year.

Acknowledgements

Allen R G, Jensen M E, Wright J L, Burman R D.1989.Operational estimates of reference evapotranspiration.Agronomy Journal, 81, 650-662.

Allen R G, Pereira L S, Raes D, Smith M.1998.Crop Evapotranspiration: Guidelines for Computing Crop Water Requirements.Irrigation and Drainage Paper No.56.vol.300.FAO, Rome.p.6541.

Abedinpour M, Sarangi A, Rajput T B S, Singh M, Pathak H,Ahmad T.2012.Performance of evaluation of AquaCrop model for maize crop in a semi-arid environment.Agricultural Water Management, 110, 55-66.

Amarasingha R P R K, Suriyagoda L D B, Marambe B, Gaydon D S, Galagedara L W, Punyawardena R, Silva G L L P,Nidumolu U, Howden M.2015.Simulation of crop and water productivity for rice (Oryza sativa L.) using APSIM under diverse agro-climatic conditions and water management technique in Sri Lanka.Agricultural Water Management,160, 132-143.

Bar-Yosef B.1999.Advances in fertigation.Advances in Agronomy, 65, 1-75.

Boogaard H L, De Wit A J W, te Roller J A, Van Diepen C A.2014.WOFOST CONTROL CENTRE 2.1; User’s guide for the WOFOST CONTROL CENTRE 2.1 and the crop growth simulation model WOFOST 7.1.7.Wageningen University and Research Centre, Wageningen, The Netherlands.

Boote K J, Jones J W, Pickering N B.1996.Potential uses and limitations of crop models.Agronomy Journal, 88, 704-716.

Bu L D, Chen X P, Li S Q, Liu J L, Zhu L, Luo S S, Hill R L,Zhao Y.2015.The effects of adapting cultivars on the water use efficiency of dryland maize (Zea mays L.) in northwest China.Agricultural Water Management, 148, 1-9.

van Bussel L G J, Grassini P, van Wart J, Wolf J, Claessens L,Yang H S, Boogaard H, de Groot H, Saito K, Cassman K G, van Ittersum M K.2015.From field to atlas: Upscaling of location-specific yield gap estimates.Field Crops Research,177, 98-108.

Claassen M M, Shaw R H.1970.Water deficit effects on corn.II.Grain Components.Agronomy Journal, 62, 652-655.

Girardin P, Tollennaar M.1994.Effects of intraspecific interactions on maize leaf azimuth.Crop Science, 34,151-155.

Goudriaan J.1986.A simple and fast numerical method for the computation of daily totals of crop photosynthesis.Agricultural and Forest Meteorology, 38, 249-254.

Grassini P, van Bussel L G J, van Wart J, Wolf J, Glaessens L, Yang H S, Boogarrd H, de Groot H, van Ittersum M K, Cassman K G.2015.How good is enough? Data requirements for reliable crop yield simulations and yieldgap analysis.Field Crops Research, 177, 49-63.

Grassini P, Yang H S, Cassman K G.2009.Limits to maize productivity in Western Corn-Belt: A simulation analysis for fully irrigated and rainfed conditions.Agricultural and Forest Meteorology, 149, 1254-1265.

Grassini P, Yang H S, Irmak S, Thorburn J, Burr C, Cassman K G.2011.High-yield irrigated maize in the Western U.S.Corn Belt: II.Irrigation management and crop water productivity.Field Crops Research, 120, 133-141.

Guan H J, Li J S, Li Y F.2013.Effects of drip system uniformity and irrigation amount on water and salt distributions in soil under arid conditions.Journal of Integrative Agriculture,12, 924-939.

He J Q, Dukes M D, Hochmuth G J, Jones J W, Graham W D.2012.Identifying irrigation and nitrogen best management practices for sweet corn production on sandy soils using CERES-Maize model.Agricultural Water Management,109, 61-70.

He Y B, Cai W M.2016.Linking a farmer crop selection model (FCS) with an agronomic model (EPIC) to simulate cropping pattern in Northeast China.Journal of Integrative Agriculture, 15, 2417-2525.

Hou P, Cui Z L, Bu L D, Yang H S, Zhang F S, Li S K.2014a.Evaluation of a modified Hybrid-Maize model incorporating a newly developed module of plastic film mulching.Crop Science, 54, 2796-2804.

Hou P, Liu Y, Xie R Z, Ming B, Ma D L, Li S K, Mei X R.2014b.Temporal and spatial variation in accumulated temperature requirements of maize.Field Crops Research, 158, 55-64.

ISS (Institute of Soil Science, Chinese Academy of Sciences).1986.The Soil Atlas of China.Cartographic Publishing House, Beijing.(in Chinese)

Irmak S, Djaman K, Rudnick D R.2016.Effects of full and limited irrigation amount and frequency on subsurface drip-irrigated maize evapotranspiration, yield, water use efficiency and yield response factors.Irrigation Science, 34, 271-286.

Jackson R D, Idso S B, Reginato R J.1981.Canopy temperature as a crop water-stress indicator.Water Resources Research, 17, 1133-1138.

Jones C A, Kiniry J R.1986.CERES-Maize: A Simulation Model of Maize Growth and Development.Texas A&M University Press, College Station, TX.

Jones H G.1992.Plant and Mocroclimate: Aquantitative Approach to Environment Plant Physiology.2nd ed.Cambridge University Press, Cambridge.

Jiang Y W, Zhang L H, Zhang B Q, He C S, Jin X, Bai X.2016.Modeling irrigation management for water conservation by DSSAT-maize model in arid northeast China.Agricultural Water Management, 177, 37-45.

Kozak J A, Ma L W, Ahuja L R, Flerchinger G, Nielsen D C.2005.Evaluating various water stress calculations in RZWQM and RZ-SHAW for corn and soybean production.Agronomy Journal, 98, 1146-1155.

Lamm F R, Trooien T P.2003.Subsurface drip irrigation for corn production: A review of 10 years of research in Kansas.Irrigation Science, 22, 195-200.

Leib B G, Jabro J D, Matthews G R.2003.Field evaluation and performance comparison of soil moisture sensors.Soil Science, 168, 396-408.

Li J R, Liu B H.2006.The change characters of monsoon rainband over Heilongjiang Province for the past 40 years.Journal of Forestry Research, 17, 71-74.

Lindquist J L, Arkebauer T J, Walters D T, Cassman K G,Dobermann A.2005.Maize radiation use efficiency under optimal growth conditions.Agronomy Journal, 97, 72-78.

Liu C, Sun B C, Tang H J, Wang T Y, Li Y, Zhang D F, Xie X Q, Shi Y S, Song Y C, Yang X H, Li J S.2017.Simple nonlinear model for the relationship between maize yield and cumulative water amount.Journal of Integrative Agriculture, 16, 858-866.

Liu S, Yang J Y, Zhang X Y, Drury C F, Reynolds W D,Hoogenboom G.2013.Modeling crop yield, soil water content and soil temperature for a soybean-maize rotation under conventional and conservation tillage systems in Northeast China.Agricultural Water Management, 123,32-44.

Liu Y, Li J, Li Y.2015.Effects of split fertigation rates on the dynamics of nitrate in soil and the yield of mulched drip-irrigated maize in the sub-humid region.Applied Engineering in Agriculture, 31, 103-117.

Liu Y, Yang H S, Li Y, Yan H J, Li J S.2017.Modeling effects of plastic film mulching on irrigated maize yield and water use efficiency in sub-humid Northeast China.International Journal of Agricultural and Biological Engineering, 10,69-84.

Liu Y, Yang S J, Li S Q, Chen F.2012.Application of the Hybrid-Maize model for limits to maize productivity analysis in a semiaird environment.Scientia Agricola, 69, 300-307.

Liu Z J, Yang X G, Hubbard K G, Lin X M.2012.Maize potential yields and yield gaps in the changing climate of northeast China.Global Change Biology, 18, 3441-3454.

Liu Z J, Yang X G, Lin X M, Hubbard K G, Lv S, Wang J.2016.Maize yield gaps caused by non-controllable, agronomic,and socioeconomic factors in a changing climate of Northeast China.Science of the Total Environment, 541,756-764.

Lizaso J I, Batchelor W D, Westgate M E, Echarte L.2003.Enhancing the ability of CERES-Maize to compute light capture.Agriculture Systems, 76, 293-311.

Maddonni G A, Otegui M E, Cirilo A G.2001.Plant population density, row spacing and hybrid effects on maize canopy architecture and light attenuation.Field Crops Research,71, 183-193.

Otegui M E.1995.Kernel set and flower synchrony within the ear of maize: II.Plant population effects.Crop Science,37, 448-455.

McMaster G S, Wilhelm W W.1997.Growing degree-days:One equation, two interpretations.Agricultural and Forest Meteorology, 87, 291-300.

Morell F, Yang H S, Cassman K G, van Wart J, Elmore R W,Licht M, Coulter J A, Ciampitti I A, Pittelkow C M, Brouder S M, Thomison P, Lauer J, Graham C, Massey R, Grassini P.2016.Can crop simulation models be used to predict local to regional maize yields and total production in the U.S.Corn Belt? Field Crops Research, 192, 1-12.

NBSC (National Bureau of Statistics of China).2015.China Statistics Yearbook.China Statistics Press, Beijing.(in Chinese)

Ritchie S W, Hanway J J, Benson G O.1992.How a Corn Plant Develops.Iowa State University, Ames.

Russel G, Marshall B, Jarvis P G.1989.Plant Canopies: Their Growth, Form and Function.Cambridge University Press,Cambridge.

Shi D Y, Li Y H, Zhang J W, Liu P, Zhao B, Dong S T.2016.Increased plant density and reduced N rate lead to more grain yield and higher resource utilization in summer maize.Journal of Integrative Agriculture, 15, 2515-2528.

Shi S Q, Cao Q W, Yao Y M, Tang H J, Yang P, Wu W B, Xu H Z,Liu J, Li Z G.2014.Influence of climate and socio-economic factors on the spatio-temporal variability of soil organic matter: A case study of central Heilongjiang Province,China.Journal of Integrative Agriculture, 13, 1486-1500.

Shirazi M A, Boersma L.1984.A unifying quantitative analysis of soil texture.Soil Science of American Journal, 48, 142-147.

Song Z W, Guo J R, Zhang Z P, Kou T J, Deng A X, Zheng C Y, Ren J, Zhang W J.2013.Impacts of planting system on soil moisture, soil temperature, and corn yield in rainfed area of Northeast China.European Journal of Agronomy,50, 66-74.

Timsina J, Jat M L, Majumdar K.2010.Rice-maize systems of South Asia: Current status, future prospects and research priorities for nutrient management.Plant and Soil, 335,65-82.

Veihmeyer F J, Hendrickson A H.1949.Methods of measuring field capacity and permanent wilting percentage of soils.Soil Science, 68, 75-94.

Wang Z, Li J, Li Y.2014.Simulation of nitrate leaching under drip system uniformities and precipitation patterns during the growing season of maize in North China Plain.Agricultural Water Management, 142, 19-28.

van Wart J, van Bussel L G J, Wolf J, Licker R, Grassini P,Nelson A, Boogaard H, Gerber J, Mueller N D, Claessens L,van Ittersum M K, Cassman K G.2013.Use of agro-climatic zones to upscale simulated crop yield potential.Field Crops Research, 143, 44-55.

Xu L, Zhang Q, Zhang J, Zhao L, Sun W, Jin Y X.2017.Extreme meteorological disasters effects on grain production in Jilin Province, China.Journal of Integrative Agriculture,16, 486-496.

Yang H S, Dobermann A, Cassman K G, Walters D T.2006.Features, applications, and limitations of the Hybrid-Maize simulation model.Agronomy Journal, 98, 737-748.

Yang H S, Dobermann A, Lindquist J L, Wa-lters D T, Arkenauer T J, Cassman K G.2004.Hybrid-maize a maize simulation model that combines two crop modeling approaches.Field Crops Research, 87, 131-154.

You L, Wood-Sichra U, Fritz S, Guo Z, See L, Koo J.2014.Spatial production allocation model (SPAM) 2005 v2.0.[2015-04-21].http://mapspam.info

Zhang J J, Li J S, Zhao B Q, Li Y T.2015.Simulation of water and nitrogen dynamics as affected by drip fertigation strategies.Journal of Integrative Agriculture, 14, 2434-2445.