Table 1. Mean physical and chemical properties of the soils used in the 2002-2004 field experiments at Changping.
Experimental treatments and design
Aerobic rice variety HD297 was grown in all three years. In 2002, four irrigation treatments (W1, W2, W3, and W4) were used in four replicates, with amounts of irrigation water varying from 152 to 374 mm (Table 2). In W1, irrigation was relatively high and uniform over the growing season, in W2 it was relatively low before PI and high after PI, in W3 it was relatively high before PI and low after PI, and in W4 it was relatively low throughout the growing season. The nitrogen (N) application was 200 kg ha–1 applied in three splits. More details of this experiment are given by Yang et al (2005). In 2003 and 2004, the experiments were laid out in a split-plot design in four replicates, with four irrigation treatments (W1, W2, W3, and W4, with the same components as in 2002) as the main plot and five N treatments as subplots. Split-plot size was 45 m2 in 2003 and 35 m2 in 2004. The amount of rainfall and irrigation water applied in the four treatments is given in Table 2. In 2003, N rates were 0 (N0), 113 (N1), 150 (N2), and 225 (N3) kg ha–1 in three splits, and 225 kg ha–1 in five splits (N4). In 2004, N rates were 0 (N0), 75 (N1), 125 (N2), 175 (N3), and 225 (N4) kg ha–1 in three splits again.
Table 2. Amount of irrigation (I) and rainfall (R) (both in mm) at different stages of aerobic rice in the 2002-04 field experiments at Changping.
W1: irrigation was relatively high and uniform over the growing season;
W2: irrigation was relatively low before PI and high after PI;
W3: irrigation was relatively high before PI and low after PI;
W4: irrigation was relatively low throughout the growing season.
Cultural practices
Chicken manure was applied in every year (except 2004) in the field with an N-P2O5-K2O equivalent of 52-45-47 kg ha–1 in early April as a general management practice for the whole experiment station.
In all years, seeds were hand-dibbled at 3-cm depth in rows 30 cm apart, at a seeding rate of 125 kg ha–1 in 2002, 135 kg ha–1 in 2003, and 120 kg ha–1 in 2004. Fertilizers P, K, Fe, and Zn were applied to obtain nonlimiting conditions (Yang et al 2005). The sowing date was 15 May 2002, 14 May 2003, and 11 May 2004, and the harvest date was 7 October 2002, 12 October 2003, and 11 October 2004. In all experiments, plots were bunded and separated by 1-m-wide strips of bare soil. The plots were kept weed-free by an application of preemergence herbicide and hand weeding after crop establishment. Pesticides were applied as appropriate for optimum crop protection. The method of irrigation was surface flooding, and irrigation water was applied through flexible hoses connected to a subsurface pressurized pipe system lifting water from a deep groundwater well.
Plant growth data
Dates of emergence, panicle initiation (PI), flowering, and physiological maturity were recorded. Crop samples were taken from two rows of 50-cm length at the early seedling stage, PI, booting, flowering, and maturity. The samples were taken in two of the four replicates in 2002-03, and in all four replicates in 2004. In 2003, all treatments were sampled, whereas only selected treatments were sampled in 2002 and 2004. Leaf surface areas were determined with a CID 201 meter in 2002 and by measuring width and length of the leaves in 2003-04. After oven drying at 70°C to constant weight, the dry weight of stems, green leaves, yellow plus dead leaves, and panicles was measured. At maturity, two 1-m row sections were harvested to determine yield components. Panicles were hand-threshed and filled grains were separated from unfilled grains by submerging them in tap water (Peng et al 2006). Grain-filling percentage was calculated as the ratio of number of filled grains over the total number of filled and unfilled grains. Grain yields were determined from harvest areas of 50 m2 in 2002 (whole plot), 3.3 m2 in 2003, and 10 m2 in 2004, expressed at 14% moisture content. Standard error (SE) and coefficient of variation (CV) were calculated for yield and total aboveground biomass at maturity for all three years and for the biomass of crop organs and leaf area index (LAI) during the growing season for 2004 only.
Soil water data
In all years, soil-water tension was measured using tensiometers installed at 20-cm depth in each irrigation treatment (except W3 in 2002). The tensiometers had a measurement range of 0–100 kPa. Soil samples were taken at 10-cm intervals from 0- to 5-cm depth to determine volumetric moisture content at different soil matric potentials by a pressure plate extractor. The amount of irrigation water applied was measured by using flow meters installed in a flexible hose.
Weather data
Daily weather data during the growing season were taken from a weather station about 5 km from our experimental site, including minimum and maximum temperature, solar radiation, wind speed, and vapor pressure. Rainfall was measured with a gauge installed in our experimental fields. Weather information during the experimental period is summarized in Table 3.
Table 3. Monthly total rainfall, monthly mean of daily temperature, and solar radiation or sunshine hours during the 2002-04 field experiments at Changping (average values from 1971 to 2000).
Simulations
Each experimental treatment in Table 2 is one corresponding simulation. The soil parameters (Table 1) and weather data (Table 3) are the same for different simulations and correspond to field conditions.
The three years’ experimental data collected near Beijing were used to parameterize and evaluate the ORYZA2000 model. We parameterized the model using one year of data and evaluated the model using the two remaining years. The evaluated variables included total aboveground biomass, biomass of crop organs, leaf area index, grain yield, and soil-water tension.
Model parameterisation
The most complete experimental data from 2003 were used to parameterize ORYZA2000 by following the procedure described in Section 2.3. Bouman and Van Laar (2006) have analyzed the root distribution of HD297 under similar experimental conditions, and showed that the maximum root depth was around 0.6 m, regardless of irrigation regime. Therefore, the maximum depth of roots was set to 0.6 m. All other common crop parameters were set to the standard values for ORYZA2000 (Bouman et al 2001).
For the two soils, the soil N supply was estimated from crop N uptake in zero-N treatments. Water retention characteristics were taken from the measured volumetric moisture contents at different soil matric potentials in the laboratory.
Biomass and LAI
Graphical examples of simulated and measured LAI and aboveground biomass, and stem, leaf, and panicle biomass, in time are given in Figure 1A for the calibration set and in Figure 1B for the validation set. In the calibration set, total biomass and LAI were underestimated at lower N levels (N0 and N1) and overestimated at higher N levels (N3 and N4). They were simulated well in the validation set. Scatter diagrams of simulated versus measured variables are given in Figure 2. Panicle biomass was overestimated in the mid-ranges in the calibration set (Fig. 2A), but underestimated in the higher ranges in the validation set (Fig. 2B).
Correlation between simulations and measurements was good for total aboveground biomass, stems, green leaves, and LAI, with RMSEn values ranging from 28% to 37% (Table 4). Overestimations of panicle biomass in the calibration set were confirmed by a low , high , and high RMSE and RMSEn. In the validation set, the performance of ORYZA2000 was better than in the calibration set, with RMSEn values ranging from 21% to 32%. In 2004, the RMSE and RMSEn were 2–3 times higher than the mean SD and CV of measured values, respectively.
Final biomass and grain yield
In the calibration set, simulated final biomass ranged from about 12,000 to 16,000 kg ha–1 only, while the measured values ranged from about 10,000 to 18,000 kg ha–1 (Fig. 3A). Consequently, the linear regression between simulated and measured values had a low slope , a high intercept , and a very low R2 (Table 5). Simulated yields were much higher than measured yields, especially in the low-value range (Fig. 3B), and there was no correlation between simulated and measured yield.
In the validation set, simulated values matched measured values much better, for both final biomass and grain yield (Fig. 3C, and 3D). The RMSE and RMSEn were about two times higher than the mean SD and CV of the measurements, respectively (Table 5).
Soil-water tension
Graphical examples of simulated and measured soil-water tension in time are given in Figure 4. In general, the temporal dynamics of soil-water tension within the 0–100 kPa range were well reproduced by ORYZA2000. In statistical terms, simulated and measured soil water agreed less than simulated and measured crop variables, as evidenced by low slopes , high intercepts , low R2, and high RMSE and RMSEn(Table 6).
References
Bouman BAM, Kropff MJ, Tuong TP, Woperies MCS, ten Berge HFM, van Laar HH. 2001. ORYZA2000: modeling lowland rice. Los Baños (Philippines): International Rice Research Institute, and Wageningen (Netherlands): Wageningen University and Research Centre. 235 p.
Bouman BAM, Van Laar HH. 2006. Description and evaluation of the rice growth model ORYZA2000 under nitrogen-limited conditions. Agric. Syst. 87:249-273.
Peng S, Buresh R, Huang J, Yang J, Zou Y, Zhong X, Wang G, Zhang F. 2006. Strategies for overcoming low agronomic nitrogen use efficiency in irrigated rice systems inChina. Field Crops Res. 96:37-47.
Yang XG, Bouman BAM, Wang HQ, Wang ZM, Zhao JF, Chen B. 2005. Performance of temperate aerobic rice under different water regimes in North China. Agric. Water Manage. 74:107-122.