APO (IRRI, Philippines)

Experimental treatment and design

Table 1 summarizes some key characteristics of the experiments. The experiments were conducted on a clay soil of the experimental farm of the International Rice Research Institute (IRRI). Improved upland cultivar Apo was grown in the dry and wet seasons from 2001 to 2003. The experiments were laid out in a split-plot design with four replicates, with water regimes in the main plot and N levels in the subplot. The water treatments were continuously flooded (FF) and aerobic with flush irrigation (AA). The N treatments were 0 and 70 kg N ha–1 in the wet season or 150 kg N and 180 kg N ha–1 in the dry season in three applications at basal and two additional topdressings. P, K, and Zn were applied as basal.

Table 1. Experimental details of the field experiments. AA is continuous aerobic conditions and FF is continuous flood conditions.

The amount of irrigation water applied was monitored in each irrigation from transplanting till maturity by using V-notch weirs. In the AA and AF plots, gauged tensiometers were installed at 15–20-cm depth and readings were made daily. In all except the AA plots, field-water levels were measured daily using 30-cm-high perforated PVC pipes installed in each plot.

The groundwater table depth was measured daily in fully perforated, 5-cm-diameter PVC pipes installed down to 1.75 or 2.0 m in the center of the bunds between the main plots.

Cultural practices

For all treatments, 29-day-old seedlings were transplanted at 2 seedlings per hill and 25 hills m–2.

Data collection

Daily weather data, including rainfall, maximum and minimum air temperature, radiation, wind speed, and relative humidity, were measured at on-site meteorological stations. Crop samples were taken at key growth stages (transplanting, mid-tillering, panicle initiation, flowering, mid-grain filling, and physiological maturity) to determine total crop biomass and leaf area index (LAI). In total, plant samples from 20 hills were taken, representing 0.50 m2. Biomass was determined after oven-drying at 70°C to constant weight. LAI was determined using a Licor LI 3100 area meter. At maturity, yield was determined from two central 5.0-m2 areas, and expressed at 14% moisture content. Soil water retention characteristics were determined in several soil layers.

Simulation design and parameterization

Following the methodology introduced in Section 2, ORYZA2000 was parameterized for Apo used in our experiments, starting with the standard crop parameters for cultivar IR72. The development rates, specific leaf area, and partitioning table were derived from our experimental data. Leaf stress parameters were parameterized by Wopereis et al (1996a), so the values of minimum and maximum soil-water tension allowed nonlimited and complete inhibition of leaf expansion. For each experiment with flooded conditions, the average percolation rate was first estimated from a water balance calculation and from daily measurements of field-water level, and then fine-tuned by model fitting (fine-tuning the parameter value until simulated field-water levels agreed best with measured field-water levels). For nonflooded conditions, the Van Genuchten equations were used to describe soil water retention and conductivity characteristics in the PADDY water balance model. The Van Genuchten parameters were calculated with the pedotransfer functions developed by Wösten et al (2001), using the measured soil texture and soil organic matter content data at the sites (Table 2). The value for the saturated conductivity (Ksat) of the least permeable layer (plow pan) was further fine-tuned by matching simulated and measured field-water levels and soil water tensions. The indigenous soil N supply was first estimated from crop N uptake in zero-N treatments, and subsequently fine-tuned by model fitting.

For each experimental year, daily groundwater depth and weather data were directly taken from the measurements.

The same numbers of simulations as field experiments were designed to validate the performance of ORYZA2000 on simulating soil water tension, field-water level, LAI, biomass, and yield.

Table 2. Soil water retention characteristics, saturated hydraulic conductivity, and parameterized Van Genuchten parameters per soil layer at Los Baños.

Model evaluation

Following the methodology in Section 2.5, the performance of ORYZA2000 in simulating soil water tension, field-water level, LAI, biomass, and yield was evaluated.

Parameter values

The parameterized soil hydraulic properties are given in Table 3. The average percolation rate of the soil varied around 5 mm d–1. The value of Ksat of the plow pan varied between 7 and 20 mm d–1, and was well within the range of values presented by Wopereis et al (1996b). High percolation rates were matched by high values of Ksat.

Table 3. Calibrated parameter values in ORYZA2000: percolation rate and saturated hydraulic conductivity of the most impermeable layer.

Soil water tension

Simulated and observed values of soil water tension are given in Figures 1A and 1B for the AA treatment in 2002. Figure 1C illustrates that the simulated and measured soil water tension values scattered away from a 1:1 relationship. The agreement between simulated and observed soil water tension was somewhat weak: R2 was lower (0.14), the slope ( = 0.64) was not close to 1, and the intercept ( = 3.99 kPa) deviated from 0. The RMSEn of soil water tension was even 182% (Table 4).

Field-water level

Figure 2 (2A and 2B) compare the course of simulated and measured field-water levels for FF in dry and wet seasons from 2001 to 2003. The varying transitions indicated by measured values were generally represented by the simulations. Figure 2C shows that the water levels were slightly overestimated, especially in the dry season. The agreement between simulated and measured field-water depth was good, and better than that for soil water tension: R2 was 0.62, the slope ( =1.00) was equal to 1, and the intercept ( = 5 mm) was close to 0. The RMSEn of soil water tension was small (44%).

LAI

Figures 3 (3A, 3B, and 3C) and 4 (4A, 4B, and 4C) illustrate the time sequential dynamics of simulated and measured LAI in dry and wet seasons for both aerobic and flooded water management practices. Regardless of the season and water management, the simulated LAI deviated relatively more from observed values with nitrogen fertilizer application than without nitrogen application. Figure 5 shows that the relationship between measured and simulated LAI was close to the 1:1 line although overestimation exists for some cases.  The agreement between measured and simulated LAI was low in both aerobic and flooded conditions because and/or values deviated more from 1 and 0, respectively, the R2 was only 0.66 for the aerobic set and 0.85 for the flooded set, and the RMSEn was relatively high (36% for the flooded set and 53% for the aerobic set) (Table 4).

Biomass

Figures 3 and 4 demonstrate how well the aboveground biomass components are simulated by the model for two nitrogen applications and two water management practices. Moreover, simulated versus measured biomass data are presented for all experiments and treatments combined with a 1:1 line and measured deviation zone in Figure 5 (5A and 5B). Most simulated values fall in the measured deviation zone, especially for total biomass and panicle biomass. The agreements between measured and simulated values are good for all treatments (Table 4).

References

Wopereis MCS, Kropff MJ, Maligaya AR, Tuong TP. 1996a. Drought-stress responses of two lowland rice cultivars to soil water status. Field Crops Res. 46:21-39.

Wopereis MCS, Bouman BAM, Tuong TP, ten Berge HFM, Kropff MJ. 1996b. ORYZA_W: rice growth model for irrigated and rainfed environments. SARP Research Proceedings, IRRI/ABDLO, Wageningen, Netherlands. 159 p.

Wösten JHM, Veerman GJ, de Groot WJM, Stolte J. 2001. Waterretentie- en doorlatendheidskarakteristieken van boven- en ondergronden in Nederland: de Staringreeks. (Water retention and conductivity characteristics of top and subsoils in The Netherlands: the Staring series). Alterra rapport 153, Wageningen Universityand Research Center. 31 p.