Data Analysis and Computer Modeling Strategy
1. Equations proposed by Campbell (1974) and Van Genuchten and Nielsen (1985) will be used for estimating soil hydraulic functions.
2. Statistical software MINITAB will be used for analyzing measured data.
3. Daily climate data observe at the CGREC will be used to drive the model.
4. The Stella software and the True BASIC language will be used to model the soil-plant system water movement (Campbell, 1985). As there will be no direct measurement of evapotranspiration, the revised water module will be tested against the field-observed soil water content data. Predicted soil water content will be compared with the long-term data accumulated at the CGREC and also will be compared with the data from the rainout shelter treatments.
5. The water module of the Hurley Pasture Model (Thornley, 1998; See Figure 1 for model diagram) will be tested. There are two important features in this module. a) Evapotranspiration is calculated with the Penman-Monteith equation that is good for describing transpiration from closed canopies. b) Only one soil layer is considered (assuming uniform soil hydraulic characteristics). This simplified treatment is certainly not compatible with the real soil situation but may be acceptable when the water module is incorporated into an ecosystem model, the complexity of the latter requires every module to be as simple as realistically possible.
6. More complexity will be added to the simple Hurley Pasture water module. a) The sparse-canopy theory proposed by Shuttleworth and Wallace (1985) in the calculation of evapotranspiration will be used. This way, evaporation from the bare soil surface of the heavily grazed pastures may be better addressed; b) Four more soil layers will be added so that water flows in the soil-root interface may be better described.
7. The behavior of the above two models (one simple and one complex) will be explored to compare the model predictions with the field measured soil water data. This way, a more desirable model (a hybrid) could be selected. The selection criteria is sketched in Figure 2. We assume first that the one-soil-layer model, because of its over-simplification, though computationally efficient (less model complexity), cannot predict the observed data well, and that a 5-layer model, for example, can provide a better prediction but it requires more computation. In exploring the behavior of the models, two possible situations are of value beginning from the one-layer model. If some parameter adjustment can significantly improve the predictive accuracy, it is desirable to keep this one-layer model; however, if parameter adjustment and adding one more layer do not improve the prediction, and only after more layers are added the model begins to show some predictive accuracy, the multi-layer model would be more desirable despite its complexity.
Table of contents | Next section | |
Methods of Field Measurement | Results for 2001 |
NDSU Central Grasslands Research Extension Center |
Home | Livestock Research | Economics Research | Range Research | 2001 Annual Report |