According to Stockle et al. (2003), the initial crop-growth simulation models, mainly theoretical approaches, appeared in the 1970s (de Wit et al., 1970; Arkin et al., 1976). Applications oriented models appeared during the 1980s (Wilkerson et al., 1983; Swaney et al., 1983). Models such as SUCROS and others associated with the Dutch 'School of de Wit' (Bouman et al., 1996); as well as those produced in the US as the CERES (Ritchie, 1998) and CROPGRO (Boote et al., 1998) families of models had a significant impact on the crop modelling community (Stockle et al., 2003). As pointed out by Brisson et al (2003), the rest of the crop-growth simulating models, although different; generally follow similar guidelines than the originally produced models. Alexandrov (2002) provided a complete summary of the crop models that have been used in Europe.
3.1.1 The DSSATModels and the "Cascade Approach"
The Decision Support System for Agrotechnology Transfer (DSSAT) was originally developed by an international network of scientists, cooperating in the International Benchmark Sites Network for Agrotechnology Transfer project (IBSNAT, 1993; Tsuji et al., 1998; Uehara, 1998; Jones et al., 1998), to facilitate the application of crop models in a systems approach to agronomic research (Jones et al., 2003).
DSSAT is a microcomputer software package that contains crop-soil simulation models, data bases for weather, soil, and crops, and strategy evaluation programs integrated with a 'shell' program which is the main user interface (Jones et al., 1998). DSSAT originally comprises the CERES models for maize (Jones and Kiniry, 1986) and wheat (Ritchie and Otter, 1985), as well as the SOYGRO soybean (Wilkerson et al., 1983) and PNUTGRO peanut (Boote et al., 1986) models, among others (Jones et al., 2003).
The decision to make these models compatible led to the design of the DSSAT and the ultimate development of compatible models for additional crops, such as potato, rice, dry beans, sunflower, and sugarcane (Hoogenboom et al., 1994; Jones et al., 1998; Hoogenboom et al., 1999; Jones et al., 2003).
According to Hoogenboom et al. (1999), the DSSAT Cropping System Model (CSM) simulates growth and development of a crop over time, as well as the soil water, carbon and nitrogen processes and management practices. The CSM main components are:
• A main driver program, which controls timing for each simulation,
• A Land unit module, which manages all simulation processes which affect a unit of land
• Primary modules that individually simulate the various processes that affect the land unit including weather, plant growth, soil processes, soil-plant-atmosphere interface and management practices.
Collectively, these components simulate the changes over time in the soil and plants that occur on a single land unit in response to weather and management practices.
DSSAT has a module format. Each module has six operational steps, (run initialization, season initialization, rate calculations, integration, daily output, and summary output). The main program controls the timing of events: the start and stop of simulation, beginning and end of crop season, as well as daily time loops (Hoogenboom et al., 1999).
Ritchie (1998) provided the background of DSSAT models regarding simulation of soil-water movement and crop water-use. The DSSAT simulation of the soil water balance depends on the capability of water from rainfall or irrigation to enter soil through the surface and be stored in the soil reserve.
The "cascading approach" as used in DSSAT is explained by Ritchie (1998). Drainage from a layer takes place only when the soil water content at a given depth is between field saturation and the drained upper limit.
The Priestley-Taylor (1972) equation for potential evapotranspiration is used in DSSAT. Calculation of potential evaporation requires an approximation of daytime temperature and the soil-plant reflection coefficient (albedo) for solar radiation. For the approximation of the daytime temperature a weighted mean of the daily maximum and minimum air temperatures is used. The combined crop and soil albedo is calculated from the model estimate of leaf area index and the input bare soil albedo (Ritchie, 1998).
The root water absorption in DSSAT is calculated using a law of the limiting approach whereby the soil resistance, the root resistance, or the atmospheric demand dominates the flow rate of water into the roots. The flow rates are calculated using assumptions of water movement to a single root and that the roots are uniformly distributed within a layer (Ritchie, 1998).
The potential transpiration and biomass production rates are reduced by multiplying their potential rates by a soil water deficit factor calculated from the ratio of the potential uptake to the potential transpiration. A second water deficit factor is calculated to account for water deficit effects on plant physiological processes that are more sensitive than the stomata controlled processes of transpiration and biomass production (Ritchie, 1998).
The DSSAT models have been indeed the most used simulation tools in agricultural climate-effect assessments (Tubiello and Ewert, 2002) and crop water balance studies (e.g. Eitzinger et al., 2002). They were calibrated and validated at many agricultural regions of the world (Hoogenboom, 2000). DSSAT models have been intensively used also in the framework of the CLIMAG activities aimed to mitigate and estimate agricultural climate-risks (Adiku et al., 2007; Meza, 2007; Singh et al., 2007).
Van Ittersum et al. (2003) provided a complete summary of the family of models made in Wageningen, The Netherlands, during the last 30 years. According to Tubiello and Ewert (2002), these Dutch models have been the most widely used, after DSSAT models, in agricultural climate-risk assessments.
As pointed out by Van Ittersum et al. (2003), the Wageningen group has a long tradition in developing and applying crop models in its agroecological research program, based on the pioneering work of C.T. de Wit. In the 1960s and 1970s the main aim of these modelling activities was to obtain understanding at the crop scale based on the underlying processes. De Wit and co-workers at the Department of Theoretical Production Ecology of Wageningen University, and the DLO Research Institute for Agrobiology and Soil Fertility developed the model BACROS and evaluated components of the model (such as canopy photosynthesis) with especially designed equipment and field experiments (De Wit et al., 1978; Goudriaan, 1977; Van Keulen, 1975; Penning de Vries et al., 1974). These modelling approaches have served as the basis and inspiration for modelling groups around the world (Stockle et al., 2003).
In the 1980s a wide range of scientists in Wageningen became involved in the development and application of crop models. The generic crop model SUCROS for the potential production situation was developed (Van Keulen et al., 1982; Van Laar et al., 1997), which formed the basis of most recent Wageningen crop models such as WOFOST (Van Keulen and Wolf, 1986), MACROS (Penning de Vries et al., 1989), and ORYZA (Bouman et al., 2001). In the 1990s the Wageningen group focused more on applications in research, agronomic practice and policy making (Van Ittersum et al., 2003).
Crop modelling in Wageningen for potential production situations follows the photosynthesis approach in the SUCROS family of models (Van Ittersum et al., 2003). LINTUL (Light INTerception and UtiLisation) models use the linear relationship between biomass production and the amount of radiation intercepted (captured) by the crop canopy (Monteith, 1981), which has been found for many crop species, grown under well-watered conditions and ample nutrient supply, in the absence of pests, diseases and weeds. This relationship sets a finite limit on yield potential (Sinclair, 1994), which thus can be modelled without going into detailed descriptions of the processes of photosynthesis and respiration. Spitters and Schapendonk (1990) developed the model LINTUL with a module for the calculation of crop growth based on the LUE concept.
In the photosynthesis approach in SUCROS (Simple and Universal CROp growth Simulator) models, the daily rate of canopy CO2 assimilation is calculated from daily incoming radiation, temperature and leaf area index (LAI). The model contains a set of subroutines that calculate the daily totals by integrating instantaneous rates of leaf CO2 assimilation (Goudriaan and Van Laar, 1994; Van Laar et al., 1997).
Particularly, the model WOFOST (WOrld FOod STudies) simulates crop production potentials as dictated by environmental conditions (soils, climate), crop characteristics and crop management (irrigation, fertiliser application) (Van Diepen et al., 1989). The model has been continuously modified, and applied for many different purposes (e.g. De Koning and Van Diepen, 1992). WOFOST uses the SUCROS approach for potential production conditions.
WOFOST permits dynamic simulation of phenological development from emergence till maturity on the basis of crop genetic properties and environmental conditions. The cultivar-
specific values of thermal time assimilate conversion coefficients, maximum rooting depth, daily root development rate and partitioning fractions are important inputs. Dry matter accumulation is estimated by the rate of gross CO2 assimilation of the canopy. This rate depends on the radiation energy intercepted by the canopy, which is a function of incoming radiation and of crop leaf area. Simulated growth processes and phenological development are regulated by temperature (e.g. the maximum rate of photosynthesis), radiation and atmospheric CO2 content and limited by availability of water. Root extension is computed in a simple way, the initial and the maximum rooting depth as determined by the crop and by the soil and the maximum daily increase in rooting depth being specified prior to the simulation. The daily increase in rooting depth is equal to the maximum daily increase unless maximum rooting depth is reached. The Ritchie (1972) equation is used to separate the evaporation and transpiration terms from the evapotranspiration.
The potential biomass production rate is assumed to decrease in the same proportion as the transpiration so that the actual amount of biomass produced on a given day and consequently during whole season can be calculated.
WOFOST has been used by the European Union's Joint Research Centre (JRC) to develop a system for regional crop state monitoring and yield forecasting for the whole European Union (Van Ittersum et al., 2003). The system comprises winter wheat, grain maize, barley, rice, sugar beet, potatoes, field beans, soybean, winter oil seed rape and sunflower. This system, called crop growth monitoring system (CGMS), generates region-specific indicators of the agricultural season conditions in the current year, on a semi-real time basis. This has been realised by simulating yields from weather and soil data, which serve as crop production indicators. This model output is qualitative in the sense that it is based on comparison of quantified indicators of the current year with those of the past. It provides information on whether in the current season a given crop deviates from the 'normal' growing pattern in terms of biomass and phenological development. These crop indicators are used in combination with regression techniques as a basis for quantitative regional yield prediction for the various crops. The system is operational for the EU and has been installed in various non-EU countries.
According to Van Ittersum et al. (2003), despite that Wageningen has a strong tradition in crop modelling, which has yielded a rich variety in crop modelling approaches and modules; there has not been a strong drive towards integration of research efforts, particularly not for implementation and application purposes.
As pointed out above, besides DSSAT and the Wageningen models, several other models have been developed during the last years, aimed to estimate crop development and yields under different agricultural management conditions. Some of these models have been developed and tested in Europe. Numerous models are now available, with different objectives, and many new models are still appearing. Actually, there is no universal model and it is necessary to adapt system definition, simulated processes and model formalisations to specific environments or to new problems (Van Ittersum et al., 2003). To efficiently manage irrigation systems has been one of the most important issues considered in simulation models since they appear.
Particularly, the European Society of Agronomy (ESA), has a special session dedicated to such modelling tools. Besides, special numbers of the European Journal of Agronomy have been dedicated to promote such models. The CROPSYST model (Stockle et al., 2003), the French model STICS (Brisson et al., 2003) and the Australian model APSIM (Keating et al., 2003) are examples of such other models.
Other available European model is the Czech model PERUN (Dubrovsky et al., 2002, 2003). The model is a computer Windows-based system for probabilistic crop yield forecasting. The system comprises all parts of the process: (1) Preparation of input parameters for crop model simulation, (2) launching the crop model simulation, (3) statistical and graphical analysis of the crop model output, (4) crop yield forecast. The weather data series are calculated by the stochastic weather generator Met&Roll (Dubrovsky, 1999) with parameters that are derived from the observed series. The synthetic weather series coherently extends the available observed series and fit the weather forecast. Wind speed and humidity were added to the standard set of four surface weather characteristics generated by Met& Roll to meet the input data requirements. These two variables are generated separately by nearest neighbours re-sampling. To prepare the weather data for seasonal crop yield forecasting, the weather generator may now generate the synthetic series which coherently follows the observed series at any day of the year.
The crop yield forecast made by PERUN is based on the WOFOST crop model simulations run with weather series consisting of observed series till DAY-1 coherently followed up by synthetic weather series since DAY. The simulation is repeated n times (new synthetic weather series are stochastically generated for each simulation) and the probabilistic forecasts are then issued in terms of the average and standard deviation of the model crop yields obtained in the n simulations. The synthetic part of the weather series is prepared by a two step.
Was this article helpful?