Modelling the effect of mulching on soil heat transfer, water movement and crop growth for ground cover rice production system

Field Crops Research 201 (2017) 97–107

Contents lists available at ScienceDirect

Field Crops Research journal homepage: www.elsevier.com/locate/fcr

Modelling the effect of mulching on soil heat transfer, water movement and crop growth for ground cover rice production system Hao Liang a , Kelin Hu a,∗ , Wei Qin b , Qiang Zuo a , Yanan Zhang a a College of Resources and Environmental Sciences, China Agricultural University, Key Laboratory of Arable Land Conservation (North China), Ministry of Agriculture, Beijing 100193, PR China b Department of Soil Quality, Wageningen UR, 6700 AA Wageningen, the Netherlands

a r t i c l e

i n f o

Article history: Received 15 April 2016 Received in revised form 11 November 2016 Accepted 11 November 2016 Keyword: Film mulching Soil temperature Water-saving Water use efficiency WHCNS model

a b s t r a c t Soil-crop system models often failed to simulate the effect of plastic film mulching (FM) on soil heat transfer, water movement and crop growth due to lack of appropriate method and the measured data in the fields. The objectives of this study were to (i) improve the Soil Water Heat Carbon Nitrogen Simulator (WHCNS) model to simulate soil temperature, water content and rice growth under FM condition, and (ii) to analyze the effect of FM on water balance and water use efficiency (WUE) under different water and nitrogen (N) management, using the data of a two-year field experiment with a factorial design of two water (Wsat and W80% , soil water content was kept at saturation and 80% field capacity) and three N levels (N1: zero-N fertilizer; N2: 150 kg urea N ha−1 ; and N3: 75 kg urea N ha−1 plus 75 kg N ha−1 as manure) treatments. The results showed that the modified model accurately simulated the changes in soil temperature, soil water content, LAI, dry matter and yield under FM condition. The normalized root mean square error (nRMSE) were 4.7%, 4.5%, 24.5%, 16.5% and 7.9%, respectively, which were significantly smaller than the results simulated by the original model. Importantly, although there were no significant differences in average crop yields between two water input levels (W80% and Wsat ), the amounts of irrigation and evaporation under W80% treatment were reduced significantly by 71.9% and 36.2%, respectively. And the WUE of W80% (1.13 kg m−3 ) was higher than that of Wsat (0.84 kg m−3 ). The ranking of WUE under different N management for W80% treatments was N2 ≈ N3 > N1. In conclusion, the modified WHCNS model performed significantly better in simulating the dynamics of water, heat, and crop growth under FM. Reduced irrigation with 80% field capacity and applying 75 kg urea N ha−1 plus 75 kg N ha−1 as manure can achieve “more yield with less water”. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Rice is the primary cereal of tropical and some temperate regions. The global rice production has been increased dramatically, from 285 million ton in 1961 to 745 million ton in 2013, due to improved cultivar, irrigation facilities, fertilization and other field management (FAO, 2016). China is the world’s largest rice production country, with a planting area of 30 million hectares which accounts for 18.7% of world’s total (FAO, 2012). Around 90% of irrigated rice in China was grown under continuous flooding or submerged soil conditions, consuming 65% of total amount of irrigation water and leading to large loss of water and thereby low water use efficiency (WUE) (Si et al., 2000). There is a large room

∗ Corresponding author. E-mail address: [email protected] (K. Hu). http://dx.doi.org/10.1016/j.fcr.2016.11.003 0378-4290/© 2016 Elsevier B.V. All rights reserved.

for achieving high rice yield with less water input (Li et al., 2007; Tao et al., 2015). Many water-saving methods had been developed to achieve the aims, such as alternative wetting-and-drying irrigation (Belder et al., 2004), dry-seeding technique (Tabbal et al., 2002), rice intensification system (Stoop et al., 2002), aerobic rice (Bouman et al., 2007) and ground cover rice production system (GCRPS) (Lin et al., 2002). Among those techniques, the GCRPS cultivation significantly helped extend rice growing areas, especially for those prone to drought or low temperature (Lin et al., 2002). High resource use efficiency in GCRPS was often considered to be related to the increased soil temperature, soil moisture and weed inhabitation (Li et al., 2007; Tao et al., 2015). Li et al. (2007) and Zhang et al. (2008) reported that GCRPS increased WUE and maintained high yield, compared to traditional flooding rice system. Furthermore, GCRPS might reduce greenhouse gas emission and have a significant advantage on water-saving, increasing the ground tem-

98

H. Liang et al. / Field Crops Research 201 (2017) 97–107

perature and preventing water body pollution (Xu et al., 2004; Gao et al., 2009). Soil temperature and soil water content are two key factors for many soil biophysical processes and crop growth (Jones and Kiniry, 1987; Hansen et al., 1990). Soil temperature can affect crop phenology, canopy development, biomass and crop yield (Stone et al., 1999). However, high soil temperature can also lead to high soil evaporation, which is considered as water loss, i.e., non-productive water to crop growth. To reduce soil evaporation, plastic film mulching (FM) is often applied in the field, which would influence soil heat transfer, soil water movement and crop growth, especially at the early stage (Li et al., 2007; Xie et al., 2005; Wang et al., 2015). The FM can effectively reduce evaporation thereby saving water and improving WUE by up to 60% (Belder et al., 2007; Qin et al., 2015). However, there were large variations between regions and crop systems. As yet, there is limited information on soil heat transfer and water movement in GCRPS because measuring these in the field is laborious and time-consuming. Furthermore, the results from the field experiments are often only relevant to a specific climate condition and/or soil type. Hence, there is a need to combine the advantages of the soil-crop model and the data of field experiments, in order to provide guidance for improving field management. Many models have been used for rice production systems (Belder et al., 2007; Feng et al., 2007; Kadiyala et al., 2015). For example, Belder et al. (2007) used the ORYZA2000 model to identify the best irrigation regime in Hubei Province, China. Feng et al. (2007) explored the options to growing rice using less water in northern China based on the ORYZA2000 model, and found that wetting-and-drying irrigation can reduce 40–70% of water input without yield loss compared with flooding irrigation. Gaydon et al. (2012) coupled the ORYZA2000 model with soil water and nutrient modules from APSIM, and the model performed equally well in simulating rice grain yield compared to the original ORYZA2000. Kadiyala et al. (2015) used DSSAT (CERES-Rice) model to develop the best management practices (BMPs) for rice-maize cropping system, the results showed that BMPs can save 41% of water input and produce 96% of the yield attainable under conventional management. Chun et al. (2016) assessed the impacts of climate change on rice yields in Southeast Asia to make recommendations for national- and farmer-level adaptation strategies appropriate to different stakeholders. These studies mainly concentrated in the flooded rice planting patterns, However, few have considered the changes of soil temperature and evaporation under FM system. Moreover, most soil-crop system models could not simulate the effect of FM on soil heat transfer, water movement and crop growth for GCRPS due to lack of quantitative method and measured field data. To quantify the effect of FM on crop growth, it is necessary to improve the existent soil-crop system models for simulation of the change of both soil temperature and soil water content under FM condition simultaneously. Han et al. (2014) modified soil heat module of DNDC (Denitrification-Decomposition) to assess the impacts of FM on regional maize yield in Northern China. However, the modified DNDC model was mainly designed to simulate soil temperature and soil water content in dryland, which is not suitable to GCRPS. Recently, an integrated soil-crop model (WHCNS, soil Water Heat Carbon Nitrogen Simulator) was developed to optimize water and N management (Hu et al., 2007; Liang et al., 2016a). The model can simulate water movement, heat transfer and nitrogen transport under a double-cropping production system in the North China Plain (Li et al., 2015a). But the model is unable to simulate the effect of mulching on soil heat transfer, water movement and crop growth under GCRPS. Therefore, the objectives of this study were to (i) improve the soil water and heat modules of WHCNS model to simulate soil

temperature, water content and crop growth in GCRPS in mountainous region of Central China, and (ii) to analyze the effect of film mulching on water balance and WUE, and identify the optimal management practice among different water and N treatments. 2. Materials and methods 2.1. Study area The experiment was conducted at a farm (32◦ 07 N, 110◦ 43 E, and 440 m ASL) in Fangxian County, which is located in the mountainous region of Hubei Province in Central China. There are two major concerns in the local rice production: seasonal water shortage and low temperatures at the beginning of the rice growth season (Tao et al., 2015). The FM has been reported as one of the most effective measures to solve these problems in this region (Lin et al., 2002; Li et al., 2007). The soil was a silt loam with a texture of 20.3% sand (0.05–2 mm), 60.0% silt (0.002–0.05 mm) and 19.8% clay ( < 0.002 mm) and in the 0–20 cm depth layer it had 21.3 g organic matter kg−1 and 1.31 g total N kg−1 . The mountainous region of Fangxian County is exposed to northern subtropical monsoon climate, with an annual mean air temperature of 14.2 ◦ C and an annual average rainfall of 830 mm. The total annual sunshine hours are 1850 ± 150 h and the frost-free period lasts 225 ± 15 days. 2.2. Experiment design and filed management The experiment was conducted over two rice growing seasons (late April-September) in 2013 and 2014. Six experimental treatments were designed (consisting of two water management combined with three N treatments). Two water treatments were: (1) Wsat , mulched with plastic film and soil water content was kept approximately saturation from transplanting until two weeks before harvest; and (2) W80% , mulched with plastic film and average soil water content in root zone (0–50 cm) was kept between 80% and 100% field capacity based on the measured soil water content every two days, water balance method is used to calculate the irrigation amount. Three N treatments were designed: (1) N1: zero-N fertilizer; (2) N2: 150 kg urea N ha−1 given as basal fertilizer; and (3) N3: 75 kg urea N ha−1 plus 75 kg N ha−1 as chicken manure, all applied as basal application. All treatments received the same amount of phosphorus (45 kg P2 O5 ha−1 as Ca(H2 PO4 )2 ) and potassium (45 kg K2 O ha−1 as KCl). All treatments were set up in a randomized block designed with three replicates, each plot was 9 m wide by 10 m long as illustrated in Fig. 1. Seedlings were transplanted on 28 April in 2013 and 2014. Harvest was done on September 10, 2013 and September 19, 2014. Details of the irrigation and fertilization can be found in elsewhere (Tao et al., 2015; Jin et al., 2016). 2.3. Observations and measurement methods Soil samples were collected from a soil profile up to 0.8 m depth for the analysis of basic physicochemical properties (Table 1). The temperature of topsoil (5 cm) was measured hourly by a thermistor sensor (EBI-20T, Ebro Instruments, Germany), soil profile temperature (10 and 20 cm depth) were measured hourly by a thermocouple sensor (CB0221, Edison state of Beijing scientific instrument co., LTD, China). The soil volumetric water content was measured every two days at 0.20 m’s interval up to 0.8 m of soil profile, using a capacitance based soil moisture sensor (Diviner 2000, Sentek, Australia). Crop dry matter and leaf area index (LAI) were measured at the stages of middle tillering, maximum tillering, panicle initiation, flowering and maturity. On each sampling date, 8 hills (0.4 m2 ) were harvested. Plant samples were washed with distilled water

H. Liang et al. / Field Crops Research 201 (2017) 97–107

99

Table 1 Soil physical and hydraulic properties for soil profile in the experimental site. Soil layer(m)

0–0.2 0.2–0.4 0.4–0.8

BD (Mg–m−3 )

1.36 1.53 1.52

Particle fraction (%)

Sand

Silt

Clay

20.2 16.7 18.6

59.9 65.4 64.4

19.7 17.8 16.9

Texture (USDA)

r (m3 m−3 )

s (m3 m−3 )

a (m−1 )

n

Ksat (cm day−1 )

Silt loam Silt loam Silt loam

0.08 0.08 0.08

0.46 0.46 0.48

0.53 0.59 0.58

1.66 1.62 1.62

6.92 0.45 0.44

Note: BD is bulk density;  r is the residual water content;  s is the saturated water content; a is the inverse of the air-entry value; n is a pore size distribution index; Ksat is the saturated hydraulic conductivity.

Fig. 1. The field layout of the plastic film mulching for each plot.

and oven-dried at 70 ◦ C to constant weight. LAI was measured as the one-sided green leaf area per unit ground surface, the green leaf area for each leaf was calculated by length × width × 0.75. At maturity, grain yield and straw dry yield were determined from a 10-m2 core plot in the center of the raised beds. Detailed information on measurements can be found in Tao et al. (2015). Meteorological data including daily rainfall, radiation, humidity, wind speed, and minimum and maximum temperatures were collected at a field meteorological station (WeatherHawk 500, Campbell Scientific, USA), which was 30 m away from the center of the experimental site. 2.4. WHCNS model The WHCNS model was used to simulate soil water movement, soil heat and N transport, and crop growth. The model integrates the key processes governing water movement and N and C cycles in the soil-crop system. These key processes include soil evaporation, crop transpiration, soil water movement, runoff, soil temperature, mineralization of fresh crop residue and soil organic N, soil inorganic N immobilization in biomass, nitrification, ammonia volatilization, denitrification and nitrous oxide emissions, crop growth. The model can be used to analyze the effects of various agricultural management practices (sowing date, planting density, straw return, crop rotation, irrigation, and fertilizer application, etc.) on water and N dynamics, organic matter turnover and crop growth (Liang et al., 2016a). In the model, the potential evapotranspiration is estimated using the Penman-Monteith method (Allen et al., 1998). The infiltration from rainfall or irrigation is computed by a modified

Green-Ampt approach (Green and Ampt, 1911). Water redistribution is simulated by the Richards equation in a soil profile in which surface evaporation and plant uptake are considered as sinks. Runoff is calculated using the approach designed by the U.S. National Resource Conservation Service (NRCS, 2004). Meanwhile, soil heat transport simulation is directly introduced from the HYDRUS-1D model (Simunek et al., 2008). Soil C and N cycling concepts are from the DAISY model (Hansen et al., 1990). The crop growth modelling was based on PS123 model (Driessen and Konijn, 1992), which is a generic dynamic crop model that can be used to simulate crop growth and development stage, LAI, biomass accumulation and allocation, maintenance respiration, growth respiration and yield formation. The water stress factor is the actual transpiration divided by the potential transpiration. The N stress is calculated based on the simulation of the crop N demand, actual soil N supply, and crop N uptake. A detailed model description is available in the literature (Hu et al., 2007; Liang et al., 2016a). The model inputs included: site location (latitude, altitude), key soil physicalchemical properties, crop data (planting and harvest date, planting density, planting depth), field management, initial soil moisture content and mineral N content, and daily meteorological data (maximum and minimum air temperature, air humidity, solar radiation, wind speed, and precipitation). 2.5. Model modification In WHCNS model, crop actual transpiration (Ta ) was determined by the potential transpiration (Tp ) and soil water content, and the ratio of Ta to Tp was adopted as the extent of water stress for crop growth module. Soil temperature is a key correction factor to soil processes (such as mineralization, ammonia volatilization, denitrification, etc.) (Liang et al., 2016a). In order to reflect the effect of FM on crop growth, it is necessary for the WHCNS model to respond to the change of soil temperature and soil water content to FM condition accurately. The simulation of one-dimensional heat transfer was taken from the HYDRUS-1D model (Simunek et al., 2008), which is described with the convection-dispersion equation:



∂T ∂ ∂T Cp () () = − Cw qT ∂t ∂z ∂z

 (1)

where T is soil temperature (◦ C), Cp () and Cw are the volumetric heat capacities (J cm−3 ◦ C −1 ) of the porous medium and the liquid phase, respectively, () is the coefficient of apparent thermal conductivity of the soil (J cm−1 d−1 ◦ C −1 ), q is the volumetric flux density (cm d−1 ). In the original version of WHCNS, the soil surface temperature was estimated by the daily air temperature and LAI (Zheng et al., 1993):

 Ttop =

 + 0.25(T  Ttop air − Ttop ) exp(−ke · LAI)

 ) (Tair ≥ Ttop

 + 0.25(T  Ttop air − Ttop )

 ) (Tair < Ttop

(2)

100

H. Liang et al. / Field Crops Research 201 (2017) 97–107

where Ttop and Ttop ’ are the temperature of the soil surface on the current day and the previous day, respectively, Tair is the average air temperature on the current day, and ke is the light extinction coefficient. And the bottom boundary temperature is estimated by: Tz0 = Taa + exp(−Z0 /De )Tam cos[2(JD − JD0 )/365 − Z0 /De ]

(3)

where TZ 0 is the soil temperature of the bottom boundary with a depth of Z0 (cm) on Julian day (JD), Taa and Tam are the annual average and amplitude of air temperature, respectively, and JD0 is the Julian day when the solar altitude is the highest (i.e., 200th and 20th for the Northern and the Southern hemispheres, respectively). De is the damping depth (cm), given by: De = [365 × 864m /(Cm )]

0.5

where Cm is the average heat capacity of the soil profile −1 ),  −1 ◦ C−1 ) of the m is the average thermal conductivity (J cm soil profile. In order to estimate the soil temperature at different depths in the soil profile, the estimated daily soil surface (Ttop ) and bottom temperatures (TZ 0 ) will be used as a variable temperature boundary condition (Dirichlet boundary condition) to solve convection-dispersion equation. To make WHCNS model applicable under FM condition, the soil heat transfer module was modified based on the method proposed by Han et al. (2014), which is a simple empirical approach with less input parameters and easy to apply in field condition. In the new module, when the plastic film is applied, the heat transfer between the soil surface and the air will be hindered. Taking into account the effect of FM on the soil surface temperature, the new soil surface temperature will be determined by the difference between the air and soil temperatures as well as by the film thickness and the thermal conductivity: dH = (T air − Ttop )/Dfilm · TCfilm · dt

(5)

dH SHsoil · dz

(6)

where dH is heat flux (J cm−2 ) from the air to the soil through the film, Tair is the air temperature (◦ C), Dfilm is the actual film thickness (5E–4 cm), TCfilm is the film thermal conductivity, set at 0.0025 J s−1 cm−1 ◦ C−1 according to Han et al. (2014), dt is the unit of time (s), Tftop is the mulched soil surface temperature (◦ C), SHsoil is the soil specific heat capacity (J cm−3 ◦ C−1 ), dz set 1 cm. In the modified model, Tftop substitutes Ttop to solve the Eq. (1). In the original model, the crop coefficient Kc (−) is used to calculate actual crop potential evaporation ETc (cm d−1 ). And combing LAI to separate potential crop transpiration Tp (cm d−1 ) and potential soil evaporation Ep (cm d−1 ) (Jones and Kiniry, 1987). In this study, we introduced a ground covering coefficient Cfilm (0–1), which represents the ratio of covering area to total ground area. Assuming surface soil is homogeneous, the Ep and Tp under FM condition can be calculated by Eqs. (7) to (10). ETc = ET0 × Kc



Ep =

(7)

ETc · exp(−0.4 × LAI)/1.1 ETc · (1 − 0.43 × LAI)

Parameters

Description

Value

Tbase Tsum Ke K-ini K-mid K-end SLA-max SLA-min AMAX R-max

Base temperature (◦ C) Accumulated temperature (◦ C) Extinction coefficient Crop coefficient in initial stage Crop coefficient in middle stage Crop coefficient in end stage Maximum specific leaf area (m2 kg−1 ) Minimum specific leaf area (m2 kg−1 ) Maximum assimilation rate (kg ha−1 h−1 ) Maximum root depth (m)

10 1750 0.6 1.0 1.45 1.0 24 12 50 0.5

(4) (J cm−3 ◦ C

Tftop = Ttop +

Table 2 Crop parameters used in WHCNS model.

LAI > 1.0 (8)

0 < LAI ≤ 1.0

Tp = ETc − Ep

(9)

Efilm = (1 − Cfilm ) · Ep

(10) (cm d−1 )

and where ET0 is the reference crop evapotranspiration calculated by the Penman-Monteith equation (Allen et al., 1998). Efilm is the potential evaporation under FM condition (cm d−1 ); according to the plastic film mulching area of the experiment field (Fig. 1), the value of Cfilm was set to 0.9 in the study (156*5/(15*6 + 156*5) = 0.90).

Actual root water uptake is calculated by Eq. (11) as:



Ta =



S(h, z)dz = Tp LR

aw (h, z)b(z)dz

(11)

LR

where Ta is actual root water uptake (or crop transpiration) (cm d−1 ), LR is root length (cm), aw (h, z) is a water stress response function (Feddes et al., 1978), and b(z) is a root distribution function (Simunek et al., 2008). 2.6. Model input parameters Liang et al. (2016b) recently analyzed the sensitivity of 18 key parameters in WHCNS model. The results showed that crop parameters had a higher sensitivity than soil hydraulic parameters, and the N transformation parameters had the lowest sensitivity among all key parameters. Hence, the order of parameter calibration was set up as following: first crop parameter, then soil hydraulic parameter, and the last N transformation parameter. The measured dataset (soil temperature, water content, LAI, crop dry matter and yield, etc.) from the treatment Wsat in 2013 was used to calibrate the model. Datasets from the W80% treatment in 2013, and from the Wsat and W80% treatments in 2014 were all used to validate the model. 2.6.1. Soil parameters Soil water retention, (h), and unsaturated hydraulic conductivity, k(h), were described using van Genuchten (1987) and Mualem (1976) models, respectively. The adjusted hydraulic parameters are shown in Table 1. Considering that the microbial activity decreases sharply with increasing soil depth, the vertical scope of microbial activity in soil was set 0 to 0.50 m in the WHCNS model, accordingly the soil N transformation processes (mineralizationimmobilization, nitrification and denitrification) are considered within this soil profile. The parameters for N transformation were derived from the default values of the Daisy model (Hansen et al., 1990), and calibrated using a model-independent optimization tools (parameter estimation, PEST) to achieve the best agreement possible between simulated LAI, crop dry matter and yield, and measured data (maximum nitrification rate, Vn ∗ , 10 g m−3 d−1 ; a half saturation constant for nitrification, Kn , 50 g m−3 ; an empirical proportionality factor for denitrification, Kd , 0.1; one-order kinetic coefficient for ammonia volatilization, Kv , 0.025). As the long-term dynamics of organic matter were not the focus of this study, organic matter decomposition parameters or decay rates were drawn from the earlier studies in China (Kröbel et al., 2010). The initial C/N ratio of residues and various organic matter distribution coefficients are given in Jensen et al. (2005). 2.6.2. Crop parameters The basic crop parameters for crop modeling listed in Table 2 were drawn from Driessen and Konijn (1992). The partition

H. Liang et al. / Field Crops Research 201 (2017) 97–107 Table 4 Model evaluation for soil temperature simulations at 5 cm depth.

Table 3 Sensitivity analysis for modified WHCNS model. Parameters

Change

ST

SWS

LAI

DM

Yield

TCfilm

20% 10% −10% −20%

3.94% 1.97% -1.97% -3.94%

<0.01% <0.01% <0.01% <0.01%

<0.01% <0.01% <0.01% <0.01%

<0.01% <0.01% <0.01% <0.01%

<0.01% <0.01% <0.01% <0.01%

Dfim

20% 10% −10% −20%

-3.28% -1.79% 1.79% 3.28%

<0.01% <0.01% <0.01% <0.01%

<0.01% <0.01% <0.01% <0.01%

<0.01% <0.01% <0.01% <0.01%

<0.01% <0.01% <0.01% <0.01%

Cfilm

20% 10% −10% −20%

-0.97% -0.17% 0.00% -0.02%

1.34% 0.78% -0.21% -0.52%

13.95% 6.58% -0.44% -8.03%

5.63% 2.97% -0.13% -2.75%

2.61% 1.67% -0.22% -1.84%

Note: ST, soil temperature; SWS, soil water storage; LAI, leaf area index; DM, dry matter.

coefficients and maximum photosynthetic activity (AMAX) were calibrated to match with measured dry matter. 2.7. Model evaluation statistics Three statistical indices were used to evaluate model performance: root mean square error (RMSE), normalized RMSE (nRMSE) and index of agreement (d) (Willmott, 1982).

  n   (Si − Mi )2  i=1

RMSE =

(12)

n

nRMSE =

RMSE

× 100

(13)

M n

d=1−

(Si − Mi )2

i=1 n

(|Si − M | + |Mi − M |)

101

(14) 2

i=1

where Si is the simulated value, Mi is the measured value, n is the ¯ is the mean of measured values. The RMSE number of values, and M represents the mean difference between observed and predicted values, while nRMSE shows the relative size of the mean difference as an unbounded percentage (Willmott, 1982). The index of agreement (0 ≤ d ≤ 1) is intended to be a descriptive measure, and is both a relative and bounded measure (Willmott, 1982). The closer the value of d is to 1, the better the model performance. 3. Results and discussion 3.1. Sensitivity analysis The detailed sensitivity analysis of all input parameters of WHCNS can be found in elsewhere (Liang et al., 2016b). In this study, we only introduced three parameters for the modified model, the film thickness (Dfilm ), the film thermal conductivity (TCfilm ) and ground covering coefficient (Cfilm ), their initial values were 5E-4 cm, 0.0025 J s−1 cm−1 ◦ C−1 and 0.8, respectively. And then, the values of Dfilm , TCfilm and Cfilm increased or decreased by 10% and 20%, respectively, to conduct the sensitivity analysis. The results of sensitivity analysis were listed in Table 3, there was a positive linear correlation between TCfilm and the output of soil temperature, Dfilm had a negative effect on soil temperature, but the change of Dfilm and TCfilm had no significant effect on soil water

Treatments

Indices

Original

Modified

LAI < 1

LAI ≥ 1

Total

LAI < 1

LAI ≥ 1

Total

Wsat

RMSE (◦ C) nRMSE (%) d

4.22 16.69 0.62

1.15 4.66 0.94

2.40 9.71 0.80

1.12 4.42 0.96

1.04 4.22 0.96

1.15 4.64 0.95

W80%

RMSE (◦ C) nRMSE (%) d

4.22 16.70 0.62

1.19 4.86 0.94

2.42 9.80 0.79

1.13 4.49 0.96

1.09 4.46 0.95

1.16 4.70 0.95

content (), LAI, crop dry matter and yield. There was a positive nonlinear correlation between Cfilm and the outputs of , LAI, crop dry matter and yield, but soil temperature showed a slightly negative nonlinear correlation with Cfilm . The soil temperature, , LAI and yield changed −0.17%, 0.78%, 6.58%, 2.97% and 1.67%, respectively, when Cfilm increased by 10%. Mulching film reduced the soil evaporation, hence increased soil water storage, which resulted in the increase of LAI (detailed in section 3.2.2 and 3.2.3). The soil temperature decreased with the increase of LAI, which led to the negative relationship between soil temperature and Cfilm (Flerchinger and Pierson, 1991). 3.2. Model performance 3.2.1. Soil surface temperature Fig. 2 shows the comparisons between the simulated and measured soil surface temperature for Wsat and W80% treatments (average of three N treatments), i.e., the simulated results were distinguished for the original and modified models. Overall, the modified model significantly improved the accuracy of soil temperature simulation with smaller RMSE and larger d, notably at the early stage of crop growth. For the stage of LAI < 1, RMSE reduced from 4.22 ◦ C to 1.13 ◦ C, by 73.3%, the average of d increased from 0.62 to 0.96, compared to the results of the original model; at the stage of LAI ≥ 1, RMSE reduced 1.17 ◦ C to 1.07 ◦ C, by 9%, the average of d increased from 0.94 to 0.96 (Table 4). For treatment Wsat , the RMSE decreased from 2.4 to 1.15, by 52.2%, nRMSE decreased from 9.7% to 4.6%, and d increased from 0.80 to 0.95 (Table 4). For treatment W80% , the values of RMSE and nRMSE of the modified model were 1.16 ◦ C and 4.70%, decreased by 52.1% and 52.0%, respectively, compared with those of original simulation. Some previous studies also showed that FM had a heating effect on soil (Li et al., 2007; Belder et al., 2007). However, the original model does not take the heating effect into account that led to lower simulated soil temperature, especially at the early stage of crop growth (Fig. 2a and b). Soil evaporation was often large at the early stage of crop growth (LAI < 1) when the canopy cover was small; at the later stage when canopy cover increased (LAI > 1), soil evaporation was relatively small (Tao et al., 2015; Wang et al., 2015). Therefore, at the later stage, the effect of FM on soil temperature decreased and crop canopy became the major factor for regulating the soil temperature (Flerchinger and Pierson, 1991). These explained large differences in simulated soil temperature between the modified and original model at the early stage (LAI < 1), but small differences at the later stage (LAI ≥ 1). 3.2.2. Soil water storage Fig. 3 shows the simulated and measured soil water storage (SWS) by original and modified models. The accuracy of SWS simulation was also improved for different treatments by the modified model, i.e., RMSE of treatment Wsat significantly decreased by 31.7% compared to the results of the original model (Table 5). The improvement of SWS simulation for treatment W80% was smaller than Wsat , RMSE decreased by 9.0%, and nRMSE decreased from 5.9%

102

H. Liang et al. / Field Crops Research 201 (2017) 97–107

Fig. 2. Comparisons of measured and simulated soil temperature at 5 cm depth for (a) Wsat and (b) W80% treatments in 2014.

Table 5 Model evaluation for soil water storage simulations. Treatments

Indices

LAI < 1

LAI ≥ 1

Total

LAI < 1

LAI ≥ 1

Total

Wsat

RMSE (mm) nRMSE (%) d

20.11 8.06 0.26

13.24 5.33 0.73

13.66 5.51 0.70

5.17 2.07 0.55

8.13 3.27 0.86

9.32 3.76 0.83

W80%

RMSE (mm) nRMSE (%) d

14.65 5.95 0.65

14.60 6.36 0.88

13.83 5.86 0.86

5.19 2.11 0.95

13.67 5.96 0.89

12.59 5.33 0.89

Original

Modified

to 5.3% (Table 5). The modified model improved the accuracy of SWS simulation, especially at the early stage of crop growth. For the stage of LAI < 1, RMSE reduced by 69.4%, the average of d increased from 0.46 to 0.75; at the stage of LAI ≥ 1, RMSE reduced by 22.5%, the average of d increased from 0.81 to 0.88 (Table 5). In this study, a ground covering coefficient Cfilm was introduced in evaporation calculations that improved the accuracy of SWS simulation overall. Due to the soil water content of Wsat kept at saturation during crop growth, the amount of evaporation was larger than that of W80% , and FM prevented higher amount of water evaporation compared with W80% , which resulted in a relatively greater improvement of SWS simulation for Wsat . The process of soil surface evaporation was affected by FM which accounted for the major component of evapotranspiration at the early stage of crop growth (Allen et al., 1998; Kang et al., 2003), therefore, there was a great improvement of SWS simulation for the modified model at this stage. The evaporation can be ignored when the crop canopy covered the ground (LAI ≥ 1) (Kang et al., 2003). The evaporation is often small at this stage and transpiration becomes the major component of evapotranspiration. Therefore, the simulated SWS by original and modified models were no significant at the stage of LAI ≥ 1. Our results clearly indicated that the modified model can be adopted to simulate soil water movement under FM condition.

Fig. 3. Comparisons of measured and simulated soil water storage under six treatments in 2014.

3.2.3. Crop growth Figs. 4 and 5 show the simulated and measured crop LAI and dry matter by original and modified models. The modified model performed significantly better in simulating LAI and dry matter,

H. Liang et al. / Field Crops Research 201 (2017) 97–107

103

14 12

(a) WsatN1

10

LAI

(b) WsatN2

Measured Orginial Modified

(c) WsatN3

8 6 4 2

0 14 12

(e) W80%N2

(d) W80%N1

(f) W80%N3

LAI

10 8 6 4 2 0

16/4

31/5 15/7 Date (d/m)

29/8

16/4

31/5

15/7 Date (d/m)

29/8

16/4

31/5

15/7 Date (d/m)

29/8

Fig. 4. Comparisons of measured and two simulated leaf area index for treatments of Wsat and W80% in 2013.

20000 Measured Orginial Modified

Dry matter (kg ha-1)

(a) WsatN1 15000

(c) WsatN3

(b) WsatN2

10000

5000 0

0 20000

Dry matter (kg ha-1)

(f) W80%N3

(e) W80%N2

(d) W80%N1 15000 10000 5000 0 16/4

31/5 15/7 Date (d/m)

29/8

16/4

31/5 15/7 Date (d/m)

29/8

16/4

31/5 15/7 Date (d/m)

29/8

Fig. 5. Comparisons of measured and two simulated crop dry matter (kg ha−1 ) for treatments of Wsat and W80% in 2013.

Table 6 Model evaluation of LAI, dry matter and yield simulations. Treatments

Indices

LAI (m2 m−2 )

Dry matter (kg ha−1 )

Yield (kg ha−1 )

Original

Modified

Original

Modified

Original

Modified

Wsat

RMSE nRMSE (%) d

2.0 44.7 0.88

0.9 19.9 0.98

2091.7 24.2 0.97

1457.6 16.9 0.99

906.5 10.6 0.84

684.7 8.0 0.86

W80%

RMSE nRMSE (%) d

3.9 69.4 0.64

1.6 29.1 0.90

4185.0 49.3 0.87

1358.2 16.0 0.99

1273.0 14.9 0.45

656.9 7.7 0.88

104

H. Liang et al. / Field Crops Research 201 (2017) 97–107

with RMSE of LAI and crop dry matter decreased by more than 50%, compared with the original model; and the values of d of the modified model were larger than 0.9, indicating good performance (Table 6). For the yield simulation, nRMSE decreased from 12.8% to 7.9% compared with the original model. FM significantly reduced soil evaporation and water stress at the early stage of crop growth (Han et al., 2014; Tao et al., 2015), which promoted crop canopy growth and enhanced crop photosynthesis, resulting in the increase of dry matter. Except for N1 treatment, the accuracy of LAI and crop dry matter simulations produced by the modified model was significantly improved for other treatments (Figs. 4 and 5). Compared with the W80% treatments, the soil water content of Wsat during crop growth showed smaller variability (Fig. 3), and FM had a relatively small impact on soil water uptake and crop water stress under saturation condition, but FM resulted in more improvement of LAI and crop dry matter simulations for W80% (Figs. 4 and 5). Figs. 4 and 5 clearly illustrated that the modified model can better simulate the accelerated crop growth and dry matter assimilation at the early stage. However, in the Wsat N1 treatment (Figs. 4 a and 5 a), there was a small difference in the simulated dry matter between the two models. Probably because there was no fertilization in the Wsat N1 treatment but soil water was kept near saturation, in such case, the N stress became the main limiting factor for crop growth since the soil accumulated N exhausted with crop growth fast at the early period of crop growth, which resulted in the higher N stress at the rest period of crop growth compared with other N treatments. 3.2.4. Model performance Previous studies suggested that an nRMSE ≤ 15% represents “good” agreement; 15–30% “moderate” agreement; and ≥30% “poor” agreement (Yang et al., 2014; Li et al., 2015b). Table 7 shows the results of calibration and validation on soil temperature, soil water storage, LAI, dry matter and yield for all treatments in the 2013 and 2014. For calibration, the values of nRMSE for soil temperature, soil water storage, LAI, dry matter and yield were 3.87%, 2.95%, 10.23%, 9.08% and 8.70%, respectively, and all the values of d > 0.85, achieving the “good” agreement. For validation, a comparison of simulated and measured soil temperature, soil water storage, dry matter and yield for all treatments in 2013 and 2014 illustrated the “good” agreement, as their statistical indices of nRMSE were ranged from 3.33% to 12.12%, the range of d was 0.74-0.99. The LAI validation for W80% in 2013 (W80% 13) and Wsat in 2014 (Wsat 14) achieved a “moderate” agreement, while a “good” agreement for W80% in 2014 (W80% 14). With the values of nRMSE for W80% 13, Wsat 14 and W80% 14 were 16.20%, 26.82% and 6.09%, respectively, and d values for W80% 13, Wsat 14 and W80% 14 were 0.89, 0.66 and 0.99, respectively. All of these indicated that the modified model was suitable for simulating soil heat transfer, water movement, crop growth for GCRPS. 3.2.5. Water consumption and WUE A comparison of simulated water balances of different treatments by original and modified models are presented in Table 8. Irrigation and rainfall were the major water inputs for GCRPS. The amounts of irrigation for Wsat and W80% accounted for 38.1% and 21.1% of total water inputs, respectively, and the rainfalls in the period of rice growth in 2013 and 2014 were 638 and 535 mm, respectively. The major water consumptions for all treatments were evapotranspiration, drainage and runoff. The two-year average evaporation rates for Wsat and W80% simulated by modified model were 119.8 and 93.3 mm, and reduced by 40.2% and 50.0%, respectively, compared with those of simulated by original model (219.3 and 222.3 mm, respectively), while the average transpiration rates for Wsat and W80% simulated by modified model were

increased by 7.7% and 23.8%, respectively, compared with those simulated by original model. Xie et al. (2005) reported that soil evaporation (Ea ) under plastic FM reduced 55% for spring maize, which was similar to our results. The amount of increased Ta was simulated by the modified model, which was related to the increase of potential transpiration with increased LAI (Fig. 4). Comparing the actual evapotranspiration (ETa ) simulated by original and modified models for different treatments, the amounts of ETa for W80% increased larger than those of Wsat (Table 8), which is probably caused by the higher water stress for crop growth under aerobic conditions (W80% ). The ETa for different treatments range from 517 mm to 631 mm, which are smaller than those reported results by other studies under non-mulching rice cultivation (Shah and Edling, 2000; Belder et al., 2007; Linquist et al., 2015). Overall, the values of Ea and ETa under FM conditions simulated by the modified model were all reduced, and the value of Ta was increased. The amounts of water drainage showed no significant difference between original and modified simulations (Table 8), the main reason was that FM reduced ETa . While the amount of runoff simulated by modified model was larger than that by original model, the reason was that FM reduced soil water infiltration. Comparing the simulation results of water consumption under different treatments, the irrigation amount for Wsat (336–390 mm) was larger than that of W80% (85–110 mm), resulted in the average amount of drainage for Wsat was about three times that of W80% . Some previous studies showed that there was a strong positive correlation between irrigation and drainage (Wang et al., 2010; Kadiyala et al., 2015; Li et al., 2015a). The simulated runoff for Wsat was larger than that of W80%, since the nearly saturated soil water content for the Wsat treatment was prone to produce more runoff (Bissonnais and Singer, 1992). The simulated water balance agreed well with the previous studies, indicating that the modified model had a capability to simulate the water balance under FM condition. Bouman et al. (2005) studied the water use of irrigated tropical aerobic rice systems in Southeast Asia, and founded that the water inputs for flooded rice cultivation was about 1500 mm, while the water inputs for aerobic rice system ranged from 790 to 1430 mm. In this study, the water input under FM condition was relative low (732–1038 mm), indicating that FM can further reduce the water input for rice cultivation system. The average WUEs for Wsat and W80% treatments were 0.84 and 1.13 kg m−3 , respectively, we concluded that maintaining 80% of the soil field capacity (W80% ) was the better way to improve WUE than the saturation cultivation (Wsat ). Li et al. (2007) compared the WUEs for rice under GCRPS in five counties in Zhejiang Province of China, and found that WUE was in the range of 0.78–1.04 kg m−3 , our result was similar to the findings. The average rice yields for treatments of Wsat and W80% , are 8334 and 8335 kg ha−1 respectively, there are no significant difference (Table 8). However, the N fertilizer treatments had a significant impact on grain yield, the average yields in two years of N1, N2 and N3 were 7367, 8808 and 8832 kg ha−1 , respectively. Jing et al. (2007) optimized the N management for rice cultivation in Southeast of China, and found that the N inputs had a positive correlation with rice yield, the highest rice yield got at fertilizer N rate of around 200 kg ha−1 . Qiao et al. (2012) found that there was no significant difference in grain yields when the N fertilizer rate was above 135 kg ha−1 in Taihu Lake region, China. In this study, the fertilization rates for N2 and N3 treatments were closer to above studies. The WUEs under same water treatment were in order of N2 ≈ N3 > N1. Considering both water drainage and WUE, we recommended reduced irrigation with 80% field capacity and applying 75 kg urea N ha−1 plus 75 kg N ha−1 as manure as the best way to achieve “more rice with less water” in the study region.

H. Liang et al. / Field Crops Research 201 (2017) 97–107

105

Table 7 Summary of the WHCNS model performance for soil temperature, soil water storage, LAI, dry matter and yield for all treatments. Items

Year

2013

Treatment

Wsat (C)

W80% (V)

2014 Wsat (V)

W80% (V)

Soil temperature (◦ C)

RMSE nRMSE (%) d

0.92 3.87 0.97

1.15 4.77 0.95

1.07 4.39 0.96

1.17 4.74 0.95

Soil water storage (mm)

RMSE nRMSE (%) d

7.43 2.95 0.85

8.16 3.33 0.85

10.73 4.62 0.90

14.55 6.28 0.91

LAI (m2 m−2 )

RMSE nRMSE (%) d

0.56 10.23 0.91

1.19 16.20 0.89

2.07 26.82 0.66

0.39 6.09 0.99

Dry matter (kg ha−1 )

RMSE nRMSE (%) d

1321 9.08 0.86

1822 11.46 0.85

621 4.47 0.99

1113 8.12 0.74

Yield (kg ha−1 )

RMSE nRMSE (%) d

643 8.70 0.93

1011 12.12 0.79

598 7.18 0.89

753 10.08 0.92

Note: C, calibration; V, validation.

Table 8 Water balance and water use efficiency simulated by the WHCNS model. Treatments

Wsat N1 M Wsat N1 O Wsat N2 M Wsat N2 O Wsat N3 M Wsat N3 O W80% N1 M W80% N1 O W80% N2 M W80% N2 O W80% N3 M W80% N3 O

2013

2014

Average

P

I

Ea

Ta

D

R

P

I

Ea

Ta

D

R

Yield

WUE

638

352

360

390

535

360

638

110

535

213

638

85

535

213

638

108

173 130 166 148 186 154 156 143 138 135 145 109

535

638

128 128 143 137 145 147 49 48 53 48 27 30

360

336

471 454 547 511 519 527 412 301 480 341 439 321

535

638

160 243 122 192 138 223 105 229 82 207 81 231

535

213

112 199 90 152 91 175 101 160 96 164 95 165

474 403 477 458 488 423 454 430 478 427 505 455

115 114 150 150 144 141 61 54 73 66 41 38

131 119 130 120 117 110 134 127 110 101 83 70

7312 5688 8827 8837 8874 8221 7423 6401 8790 8751 8791 8622

0.74 0.6 0.91 0.95 0.86 0.85 0.99 0.86 1.22 1.19 1.18 1.15

Note: M, modified; O, original; Ea , evaporation (mm); Ta , transpiration (mm); D, drainage (mm); R, runoff (mm); P, precipitation (mm); I, irrigation (mm); WUE, water use efficiency (Yield/(P + I), kg m−3 ).

3.3. Model limitations

4. Conclusion

Considering the energy balance method is so complex and needs more inputs under FM condition, in this study, we proposed a simple empirical approach with less input parameters, and the results showed that the simple empirical approach is acceptable in our study area. However, when surface soil temperature is much higher than air temperature in a very dry area, an energy balance approach including the impact of radiation may improve the model. Dong et al. (2014) reported that the FM increased the soil temperature and speeded out crop emergence. However, in our study the rice seedlings were transplanted, so the impact of soil temperature change was not simulated. Moreover, some studies reported that the root growth under FM condition was different from nonmulching condition (Maurya and Lal, 1981; Kumar and Dey, 2011; Ning et al., 2015). In this study, the modified model ignored the effect of FM on crop root growth. In addition, soil N turnover and the pathway of N loss were also different under the GCRPS compared to flooded rice cultivation (Liu et al., 2005; Fan et al., 2005; Wang et al., 2015). GCRPS can effectively reduce greenhouse gas emission had also been reported (Yao et al., 2014). Currently, the WHCNS model was limited in testing the effect of FM on crop root growth, soil N cycle and N balance, which all need further studies.

The modified WHCNS model performed well in simulating soil temperature, soil water content, LAI, dry matter and crop yield, their root mean square errors (RMSE) simulated by the modified model reduced by 52.1%, 20.4%, 57.0%, 48.9% and 36.4%, respectively, compared with the results obtained by the original model; the values of nRMSE were also significantly smaller than the results of original simulations. It clearly indicated that the modified model can robustly simulate the effect of FM on soil heat transfer, water movement and crop growth in GCRPS. There were no significant differences in crop yields between Wsat and W80% treatments. The amount of irrigation for W80% was remarkably reduced by 71.9%, the amounts of water drainage and runoff were reduced by 69.2% and 13.7%, respectively. And the WUE of W80% (1.13 kg m−3 ) was larger than that of Wsat (0.84 kg m−3 ). The ranking of WUE under different N management for W80% treatments was N2 ≈ N3> N1. Our findings suggest that maintaining 80% of the field capacity and applying 75 kg N ha−1 chemical fertilization plus 75 kg N ha−1 as manure is the best way to achieve “more rice with less water”.

106

H. Liang et al. / Field Crops Research 201 (2017) 97–107

Acknowledgements This work is supported by the National Natural Science Foundation of China (No. 51139006) and Program for Changjiang Scholars and Innovative Research Team in University (IRT0412). We thank Dr. Xingxing Jin for providing soil temperature and water content data for our study. References Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration Guidelines for computing crop water requirements- FAO Irrigation and Drainage Paper No. 56. Rome. Belder, P., Bouman, B., Cabangon, R., Guoan, L., Quilang, E., Yuanhua, L., Spiertz, J., Tuong, T., 2004. Effect of water–saving irrigation on rice yield and water use in typical lowland conditions in Asia. Agric. Water Manage. 65, 193–210. Belder, P., Bouman, B.A.M., Spiertz, J.H.J., 2007. Exploring options for water savings in lowland rice using a modelling approach. Agric. Syst. 92 (1), 91–114. Bissonnais, Y.L., Singer, M.J., 1992. Crusting, runoff, and erosion response to soil water content and successive rainfalls. Soil Sci. Soc. Am. J. 56 (6), 1898–1903. Bouman, B.A.M., Peng, S., Castaneda, A.R., Visperas, R.M., 2005. Yield and water use of irrigated tropical aerobic rice systems. Agric. Water Manage. 74 (2), 87–105. Bouman, B., Feng, L., Tuong, T., Lu, G., Wang, H., Feng, Y., 2007. Exploring options to grow rice using less water in northern China using a modelling approach II. Quantifying yield, water balance components, and water productivity. Agric. Water Manage. 88, 23–33. Chun, J.A., Li, S., Wang, Q., Lee, E.J., Horstmann, N., Park, H., Veasna, T., Vanndy, L., Pros, K., Vang, S., 2016. Assessing rice productivity and adaptation strategies for Southeast Asia under climate change through multi-scale crop modeling. Agric. Syst. 143, 14–21. Dong, B., Liu, M., Jiang, J., Shi, C., Wang, X., Qiao, Y., Liu, Y., Zhao, Z., Li, D., Sia, F., 2014. Growth, grain yield, and water use efficiency of rain-fed spring hybrid millet (setaria italica) in plastic-mulched and un-mulched fields. Agric. Water Manage. 143 (9), 93–101. Driessen, P.M., Konijn, N.T., 1992. Land Use System Analysis. Wageningen Agricultural University, Wageningen, Netherlands. FAO, 2012. http://faostat.fao.org/site/339/default.aspx. FAO, 2016. http://faostat3.fao.org/home/E. Fan, M., Jiang, R., Liu, X., Zhang, F., Lu, S., Zeng, X., Christie, P., 2005. Interactions between non-flooded mulching cultivation and varying nitrogen inputs in rice–wheat rotations. Field Crop Res. 91 (2), 307–318. Feddes, R.A., Kowalik, P.J., Zaradny, H., 1978. Simulation of field water use and crop yield. Centre for Agricultural Publishing and Documentation, Wageningen. Feng, L., Bouman, B.A.M., Tuong, T.P., Cabangon, R.J., Li, Y., Lu, G., Feng, Y., 2007. Exploring options to grow rice using less water in northern China using a modelling approach: I. Field experiments and model evaluation. Agric. Water Manage. 88 (1), 1–13. Flerchinger, G.N., Pierson, F.B., 1991. Modeling plant canopy effects on variability of soil temperature and water. Agric. For. Meteorol. 56 (3), 227–246. Gao, Y., Li, Y., Zhang, J., Liu, W., Dang, Z., Cao, W., Qiang, Q., 2009. Effects of mulch, N fertilizer, and plant density on wheat yield, wheat nitrogen uptake, and residual soil nitrate in a dryland area of China. Nutr. Cycl. Agroecosyst. 85 (2), 109–121. Gaydon, D.S., Probert, M.E., Buresh, R.J., Meinke, H., Suriadi, A., Dobermann, A., Bouman, B., Timsina, J., 2012. Rice in cropping systems–Modelling transitions between flooded and non-flooded soil environments. Eur. J. Agron. 39, 9–24. Green, W.H., Ampt, G.A., 1911. Studies on soil physics: part 1. The flow of air and water through soils. J. Agric. Sci. 4 (1), 1–24. Han, J., Jia, Z., Wu, W., Li, C., Han, Q., Zhang, J., 2014. Modeling impacts of film mulching on rainfed crop yield in Northern China with DNDC. Field Crop. Res. 155, 202–212. Hansen, S., Jensen, H.E., Nielsen, N.E., Svendsen, H., 1990. NPo-research, A10: Daisy: Soil Plant Atmosphere System Model. The National Agency of Environmental Protection, Copenhagen. Hu, K.L., Li, B.G., Chen, Y., Guo, Y.Q., 2007. Coupled simulation on crop growth and soil water-heat-nitrogen transport I. Model. J. Hydraul. Eng. 38 (7), 779–785 (in Chinese with English abstract). Jensen, L.S., Salo, T., Palmason, F., Breland, T.A., Henriksen, T.M., Stenberg, B., Pedersen, A., Lundstrom, C., Esala, M., 2005. Influence of biochemical quality on C and N mineralisation from a broad variety of plant materials in soil. Plant Soil. 273 (1–2), 307–326. Jin, X., Zuo, Q., Ma, W., Li, S., Shi, J., Tao, Y., Zhang, Y., Liu, Y., Liu, X., Lin, S., Ben-Gal, A., 2016. Water consumption and water-saving characteristics of a ground cover rice production system. J. Hydrol. 540, 220–231. Jing, Q., Bouman, B.A.M., Hengsdijk, H., Van Keulen, H., Cao, W., 2007. Exploring options to combine high yields with high nitrogen use efficiencies in irrigated rice in China. Eur. J. Agron. 26 (2), 166–177. Jones, C.A., Kiniry, J.R., 1987. CERES-Maize: A Simulation Model of Maize Growth and Development. College Station: Texas A & M University Press. Kadiyala, M.D.M., Jones, J.W., Mylavarapu, R.S., Li, Y.C., Reddy, M.D., 2015. Identifying irrigation and nitrogen best management practices for aerobic rice-maize cropping system for semi-arid tropics using CERES-rice and maize models. Agric. Water Manage. 149, 23–32.

Kang, S., Gu, B., Du, T., Zhang, J., 2003. Crop coefficient and ratio of transpiration to evapotranspiration of winter wheat and maize in a semi-humid region. Agric. Water Manage. 59 (3), 239–254. Kröbel, R., Sun, Q.P., Ingwersen, J., Chen, X.P., Zhang, F.S., M"1 ller, T., Römheld, V., 2010. Modelling water dynamics with DNDC and DAISY in a soil of the North China Plain: a comparative study. Environ. Modell. Softw. 25 (4), 583–601. Kumar, S., Dey, P., 2011. Effects of different mulches and irrigation methods on root growth, nutrient uptake, water-use efficiency and yield of strawberry. Sci. Hortic. Amsterdam 127 (3), 318–324. Li, Y.S., Wu, L.H., Zhao, L.M., Lu, X.H., Fan, Q.L., Zhang, F.S., 2007. Influence of continuous plastic film mulching on yield, water use efficiency and soil properties of rice fields under non-flooding condition. Soil. Till. Res. 93 (2), 370–378. Li, Z.J., Hu, K.L., Li, B.G., He, M.R., Zhang, J.W., 2015a. Evaluation of water and nitrogen use efficiencies in a double cropping system under different integrated management practices based on a model approach. Agric. Water Manage. 159, 19–34. Li, Z.T., Yang, J.Y., Drury, C.F., Hoogenboom, G., 2015b. Evaluation of the DSSAT-CSM for simulating yield and soil organic C and N of a long-term maize and wheat rotation experiment in the Loess Plateau of Northwestern China. Agric. Syst. 135, 90–104. Liang, H., Hu, K.L., Batchelor, W.D., Qi, Z.M., Li, B.G., 2016a. An integrated soil-crop system model used for water and nitrogen management in North China. Sci. Rep. 6, 25755 http://dx.doi.org/10.1038/srep25755. Liang, H., Hu, K.L., Li, B.G., 2016b. Parameter optimization and sensitivity analysis of soil-crop system model using PEST. Trans. Chin. Soc. Agric. Eng. 32 (3), 78–85 (in Chinese with English abstract). Lin, S., Dittert, K., Tao, H., Kreye, C., Xu, Y., Shen, Q., Fan, X., Sattelmacher, B., 2002. The ground-cover rice production system (GCRPS): a successful new approach to save water and increase nitrogen fertilizer efficiency? In: Bouman, B.A.M., Hengsdijk, H., Hardy, B., Bindraban, P.S., Tuong, T.P., Ladha, J.K. (Eds.), Water-wise Rice Production: Proceedings of the International Workshop on ˜ ˜ Philippines. 8–11 Apr. IRRI, Los Banos, Water-wise Rice Production, Los Banos, Philippines, pp. 187–196. Linquist, B., Snyder, R., Anderson, F., Espino, L., Inglese, G., Marras, S., et al., 2015. Water balances and evapotranspiration in water-and dry-seeded rice systems. Irrig. Sci. 33 (5), 375–385. Liu, X., Ai, Y., Zhang, F., Lu, S., Zeng, X., Fan, M., 2005. Crop production, nitrogen recovery and water use efficiency in rice–wheat rotation as affected by non-flooded mulching cultivation (NFMC). Nutr. Cycl. Agroecosyst. 71 (3), 289–299. Maurya, P.R., Lal, R., 1981. Effects of different mulch materials on soil properties and on the root growth and yield of maize (Zea mays) and cowpea (Vigna unguiculata). Field Crop Res. 4, 33–45. Mualem, Y., 1976. A new model for predicting the hydraulic conductivity of unsaturated porous media. Water Resour. Res. 12 (3), 513–522. NRCS, 2004. Estimation of direct runoff storm rainfall. In: National Engineering Hand Book, Part630-Hydrology. USDA Washington D.C. Ning, S., Shi, J., Zuo, Q., Wang, S., Ben-Gal, A., 2015. Generalization of the root length density distribution of cotton under film mulched drip irrigation. Field Crop. Res. 177, 125–136. Qiao, J., Yang, L., Yan, T., Xue, F., Zhao, D., 2012. Nitrogen fertilizer reduction in rice production for two consecutive years in the Taihu Lake area. Agric. Ecosyst. Environ. 146 (1), 103–112. Qin, W., Hu, C.S., Oenema, O., 2015. Soil mulching significantly enhances yields and water and nitrogen use efficiencies of maize and wheat: a meta-analysis. Sci. Rep. UK 5. Shah, S.B., Edling, R.J., 2000. Daily evapotranspiration prediction from Louisiana flooded rice field. J. Irrig. Drain. E-ASCE 126 (1), 8–13. Si, T.S., Wang, H.Z., Zhang, W., 2000. Discussion and suggestion on some rive-saving problems in China. Irrig. Drain. 19 (1), 30–33 (in Chinese with English abstract). Simunek, J., Saito, H., Sejna, M., van Genuchten, M.T., 2008. The HYDRUS-1D Software Package for Simulating the One-dimensional Movement of Water, Heat, and Multiple Solutes in Variably Saturated Media. Version 4.0. U.S. Salinity Laboratory, Agricultural Research Service. U.S. Department of Agriculture, Colorado School of Mines Publishers, Riverside, CA. Stone, P.J., Sorensen, I.B., Jamieson, P.D., 1999. Effect of soil temperature on phenology, canopy development, biomass and yield of maize in a cool-temperate climate. Field Crop Res. 63 (2), 169–178. Stoop, W.A., Uphoff, N., Kassam, A., 2002. A review of agricultural research issues raised by the system of rice intensification (SRI) from Madagascar: opportunities for improving farming systems for resource–poor farmers. Agric. Syst. 71, 249–274. Tabbal, D., Bhuiyan, S., Sibayan, E., 2002. The dry–seeding technique for saving water in irrigated rice production systems. Direct seeding in Asian rice systems: strategic research issues and opportunities. Proceedings of an International Workshop, on Direct Seeding in Asia, Bangkok, Thailand, 25–28 January 2000, 281–297. Tao, Y., Zhang, Y., Jin, X., Saiz, G., Jing, R., Guo, L., et al., 2015. More rice with less water-evaluation of yield and resource use efficiency in ground cover rice production system with transplanting. Eur. J. Agron. 68, 13–21. Wang, H.Y., Ju, X.T., Wei, Y.P., Li, B.G., Zhao, L.L., Hu, K.L., 2010. Simulation of bromide and nitrate leaching under heavy rainfall and high-intensity irrigation rates in North China Plain. Agric. Water Manage. 97, 1646–1654.

H. Liang et al. / Field Crops Research 201 (2017) 97–107 Wang, X.K., Li, Z.B., Xing, Y.Y., 2015. Effects of mulching and nitrogen on soil temperature water content, nitrate-N content and maize yield in the Loess Plateau of China. Agric. Water Manage. 161, 53–64. Willmott, C.J., 1982. Some comments on the evaluation of model performance. Bull. Am. Meteorol. Soc. 63, 1309–1313. Xie, Z.K., Wang, Y.J., Li, F.M., 2005. Effect of plastic mulching on soil water use and spring wheat yield in arid region of northwest China. Agric. Water Manage. 75 (1), 71–83. Xu, Y.C., Shen, Q.R., Li, M.L., Dittert, K., Sattelmacher, B., 2004. Effect of soil water status and mulching on N2 O and CH4 emission from lowland rice field in China. Biol. Fert. Soils 39 (3), 215–217. Yang, J.M., Yang, J.Y., Liu, S., Hoogenboom, G., 2014. An evaluation of the statistical methods for testing the performance of crop models with observed data. Agric. Syst. 127, 81–89.

107

Yao, Z.S., Du, Y.B., Tao, Y.Y., Zheng, X.H., Lin, S., Butterbach-Bahl, K., 2014. Water–saving ground cover rice production system reduces net greenhouse gas fluxes in an annual rice–based cropping system. Biogeosciences 11, 6221–6236. Zhang, Z., Zhang, S., Yang, J., Zhang, J., 2008. Yield, grain quality and water use efficiency of rice under non-flooded mulching cultivation. Field Crop Res. 108 (1), 71–81. Zheng, D., Hunt, E.R., Running, S.W., 1993. A daily soil temperature model based on air temperature and precipitation for continental applications. Clim. Res. 2, 183–191. van Genuchten, M.T., 1987. A Numerical Model for Water and Solute Movement in and Below the Root Zone. US Salinity Laboratory, Riverside, CA.