Forcing functions

Exercise objective:

·    Overruling simulated with observed values: forcing functions.

Suggested reading:

Chapters 2.3, 4.1, 7.5 of the book, ORYZA2000: modelling lowland rice.

Exercise: 

Two of the most important factors that determine growth are the amount of intercepted light (photosynthetically active radiation, i.e., the wavelength between 400 and 700 nm) and the efficiency with which this light is used in photosynthesis. For a given amount of sunlight, the light interception by the crop is determined by the leaf area index (LAI). The efficiency with which this light is used in photosynthesis depends on the leaf nitrogen content (NFLV). When we have experimental data on LAI and NFLV from a specific crop, we can use this information to improve the simulation with ORYZA2000. This is especially useful when we are not sure whether the growth conditions of the field experiment we want to study were actually truly ‘potential’ (for example, the crop in the experiment may have suffered from lack of nutrients or from diseases).

Ex-II.12. In CONTROL.DAT, change the experimental data file to IR72DS0.T92. This data file contains the management information and crop measurements of the zero nitrogen treatment of the field experiment at IRRI in 1992 (see Box II.1). The information available from this experiment in Box II.1 does not allow us to run ORYZA2000 with a nitrogen balance, so the RUNMOD is still ‘POTENTIAL’. Run the model, and make graphs of the simulated and observed LAI versus time (Figure II.6a) and of the simulated and observed weights of total biomass (Figure II.6b) and panicle biomass. Explain the results. A. With no N fertilizer, the crop suffers from nitrogen deficiency and crop performance is below potential, both in terms of leaf area and biomass.

Ex-II.13. Locate the parameter NFLV_FRC in the file IR72DS0.T92 and notice its meaning: when its value is 0, simulated values (i.e. values read from the crop data file IR72.DAT; see the text between Exercises III.3 and III.4 in “Chapter III: Nitrogen-limited production”) are used, and when its value is 2, the measured data are used (‘forced’ to replace the simulated values). Locate also the parameter LAI_FRC and notice the same meaning for the use of simulated or measured LAI values. Make a rerun with the values of both parameters set to 2:

NFLV_FRC = 2

LAI_FRC = 2

Run ORYZA2000 and make graphs of simulated and observed LAI (Figure II.7a), simulated and observed total biomass (Figure II.7b), simulated and observed green leaf biomass (Figure II.7c) and simulated and observed panicle biomass, for both the default run (no forcing) and the rerun (forcing). Explain the differences. A: with forced values for LAI and NFLV, the match between observed and simulated biomass is better than without using forced values.

In Figure II.7a, we see that the simulated LAI in the forcing mode (rerun) is an interpolation of the measured LAI. However, after the last observed value of LAI (after daynumber 110), simulated LAI shoots up from less than 1 to around 1.5. This is because when the forcing mode is “on”, interpolated values are used from the first till the last observed value only. Before and after the last observed value, simulations are used as if there were no observations. Hence, pretty large “jumps” may occur in simulated values before and after the last observations.

Ex-II.14. In the RERUNS.DAT, insert an * in front of the line NFLV_FRC = 2, run ORYZA2000 and examine the results by producing the same graphs as in EX-II.13. Then remove this *, insert an * in front of the line LAI_FRC = 2, run ORYZA2000 and examine the results by generating the graphs. Explain the differences. A. The effect of LAI is most important, and the effect of leaf nitrogen content on photosynthesis is minor. We will see later (“Chapter III: Nitrogen-limited production”) that the reduced leaf area development in the zero N treatment is caused primarily by a reduced relative leaf growth rate (RGRLMAX) caused by (early) N limitations (see also Chapter 5 of the book “ORYZA2000: modelling lowland rice”).

Figure II.6a. Simulated and observed leaf area index (-) versus time, for the zero N treatment at IRRI, Los Baños, Philippines, dry season 1992.

Figure II.6b. Simulated and observed total biomass and panicle biomass (kg ha-1) versus time, for the zero N treatment at IRRI, Los Baños, Philippines, dry season 1992.

 Figure II.7a. Simulated and observed leaf area index (-) versus time for the zero N treatment at IRRI, Los Baños, Philippines, dry season 1992, with and without forcing of measured LAI and NFLV.

 Figure II.7b. Simulated and observed total biomass (kg ha-1) versus time for the zero N treatment at IRRI, Los Baños, Philippines, dry season 1992, with and without forcing of measured LAI and NFLV.

Figure II.7c. Simulated and observed green leaf biomass (kg ha-1) versus time for the zero N treatment at IRRI, Los Baños, Philippines, dry season 1992, with and without forcing of measured LAI and NFLV.