XD90247 and HD297 (Kaifeng, China)

Table 1. Key soil properties at Gaozhai Village (2001 experiment site) and at Panlou Village (2002-2004 experiment sites).

Description of experiment

In 2001-02, lowland rice variety XD90247 was used, and 37-day-old seedlings were transplanted into puddled soil at 6 plants per hill in 20 × 20-cm spacing. In 2003-04, aerobic rice variety HD297 was manually sown at 30 cm between rows with 150 kg seed ha–1. All experiments were kept as free from weeds, pests, and diseases as possible.  Irrigation water was applied using flexible hoses.

In 2001, a randomized block design was used in three replicates, with three water treatments: continuous flooding (CF) in puddled soil, alternate wetting and drying (AWD) in puddled soil, and flush irrigation (FI) at –50 kPa soil-water tension at 15-cm depth, in nonpuddled, aerobic soil.

In 2002, there were four replicates and four water treatments in nonpuddled, aerobic soil: flush irrigation at 10, 30, and 70 kPa soil-water tension at 15-cm depth, and one partially rainfed (RF) treatment with survival irrigation only when the rice crop showed very severe drought symptoms. In both 2001 and 2002, the water treatments were implemented after a 10–17-day period in which the fields were kept flooded to promote recovery from transplanting. In both years, a total of 180 kg fertilizer N ha–1 was applied in three splits.

In 2003 and 2004, a split-plot design was used with water as the main treatment and nitrogen as the subtreatment. The water treatments were flush irrigation at 30 kPa soil-water tension at 15-cm depth, and partially rainfed with survival irrigation (SI). The nitrogen treatments were 225 and 300 kg fertilizer N ha–1, both applied in five splits.

Data collection

Dates of crop development stages were recorded at emergence, transplanting (only in 2001-02), panicle initiation, and physiological maturity. Some 5–7 times during the growing season, crop samples were taken to measure leaf area index and the dry weights of green leaves, brown/yellow leaves, stems, and panicles. At maturity, 6 m2 were harvested to determine grain yield at 14% moisture content. Yield components such as grains per panicle and 1,000-grain weight were determined from a subsample at harvest. Daily observations were made of ponded water depth (in the continuous flooding treatment only), soil water potential at 15-cm depth using gauged tensiometers, percolation rate using 40-cm-long percolation rings that were closed at the top to prevent evaporation, and groundwater depth. Irrigation water was measured with flow meters installed in flexible hoses. Soil samples were taken down to 100-cm depth to measure texture, bulk density, soil organic matter content, saturated conductivity, and water conductivity and retention characteristics.

Daily values of rainfall, pan evaporation, sunshine hours, maximum and minimum temperature, and wind speed were collected from the meteorological station at the Henan experiment station, some 8 km away from the site in 2001 and 1 km from the site in 2002-04. A detailed description of the experiments and the results obtained in 2001 and 2002 is presented by Cabangon et al (2003).

Simulation design

The simulations were designed to mimic the field experiments, and the measured and recorded soil and climate data were used to feed all these simulations (Tables 2 and 3). The cultivation practice of a simulation is designed to be the same as the experimental treatment.

Simulation parameterization

ORYZA2000 was parameterized by following the procedure described in Section 2. The data of experiments in both 2001 and 2002 were used for XD90247, and the data of experiments in 2003 and 2004 were used for HD297.

The average daily percolation rate in 2001 puddled soil (for use in PADDY) was first computed from the measurements, and then fine-tuned by model fitting (simple calibration). The Van Genuchten parameters (for use in both PADDY and SAWAH) were derived from the measured soil water conductivity and retention data. For the 2001 puddled soil, the measured saturated conductivity of the plow sole was fine-tuned by matching simulated and measured soil-water tension values. The same was done for the saturated conductivity in the root zone of the nonpuddled aerobic soils. For each experimental year, daily groundwater depth and weather data were taken directly from the measurements.

Simulation evaluation

The performance of ORYZA2000 for XD90247 and HD297 under lowland and aerobic soil conditions was evaluated using all four years of experimental data. Since two years of experimental data were needed for each variety to arrive at a good parameterization, we could not distinguish an independent “validation” data set. The evaluation was undertaken by following the procedure introduced in Section 2.

Parameter values

Tables 2 and 3 present the calibrated development rates of rice phenology for varieties XD90247, and HD297.

Table 2. The phenology development rate for transplanted XD90247.

Table 3. The phenology development rate for transplanted HD297.

Ponded-water depth

Ponded-water level was measured only in the XD90247 field.  The agreement between simulated and observed field-water depth was strong: R2 was significant, the slope was close to 1, and the intercept did not deviate from 0 (Table 4).  However, the RMSEn was slightly large (50%).

Soil-water tension

The simulated and observed values of soil-water tension are given in Figure 1 for the HD297 fields in 2003-04 and in Figure 2 for XD20947 fields in 2001-02. Both graphs show that simulated soil-water tension represents measured dynamics. The agreement between simulated and observed soil-water tension was somewhat weak, especially for the HD297 field: R2 was lower, the slope was not close to 1, and the intercept deviated from 0 (Table 4). The RMSEn of soil-water tension was even 52% for HD297 fields and 81% for XD20947 fields.

Leaf area index (LAI)

Figures 3 and 4 illustrate the simulated and measured leaf area index (LAI) in different experimental treatments with rice varieties HD297 and XD20947. The simulations represent the temporal dynamics of LAI for these two rice varieties under different water and fertilizer management practices. The agreement between simulated and measured LAI was excellent: R2 was significant (0.75-0.91), and values were close to 1 and 0, respectively, and the RMSEn was less than 17% (Table 5).

Nitrogen uptake

Figures 5 (5A and 5B) and 6 illustrate the simulated and measured nitrogen contents in the aboveground organs of crops of different experimental treatments with rice varieties HD297 and XD20947. The simulations generally represent the dynamics of nitrogen content in aboveground organs. However, remarkable underestimations occurred in total nitrogen contents for all treatments in fields with HD297 in 2001 and 2002. These underestimations mainly resulted from the underestimations of stem nitrogen contents. For the experimental treatments with XD20947, the measured nitrogen contents were well represented by simulations. However, the obvious underestimations of grain nitrogen contents in the maturity stage resulted in underestimations of total nitrogen contents. The agreement between simulated and measured nitrogen uptake at harvest was good for all experimental treatments of both rice varieties: R2 was significant (0.91), values were close to 0, the RMSEn was less than 17%, and the value for rice XD20947 was close to 1, but not for HD297 (Table 5).

Biomass

A comparison between the course of simulated and measured biomass of aboveground organs is presented in Figure 7 (7A and 7B) for variety HD297 in 2003 and 2004, and in Figure 8 for variety XD20947 in 2001 and 2002. The measured temporal dynamics in total aboveground biomass and stem, leaf, and panicle biomass were represented by the model. The agreements between simulated and measured biomass were excellent: the values of R2 were higher than 0.82, the values of were close to 1, and the values of were low except for the total biomass of HD297 (Table 5). Moreover, the values of RMSEn of biomass were lower than 41%.

Grain yield

Table 5 presents the evaluation results for grain yields of both varieties. The agreement between simulated and measured yields of XD20947 was excellent: R2 was 0.94, equaled 1, was very small, and the RMSEn was only 11%. The agreement for HD297 was weaker than for XD20947.

Reference

Cabangon, R., Lu, G., Tuong, T.P., Bouman, B.A.M., Feng, Y., Zhang, Z., 2003. Irrigation management effects on yield and water productivity of inbred and aerobic rice varieties in Kaefeng. In: Proceedings of the First International Yellow River Forum on River Basin Management, vol. 2. The Yellow River Conservancy Publishing House,Zhengzhou, Henan, China, pp. 65–76.