The CropSyst Model: A brief description

Marcello Donatelli's home page

Introduction
CropSyst (Cropping Systems Simulation Model) is a multi-year, multi-crop, daily time step crop growth simulation model, developed with emphasis on a friendly user interface, and with a link to GIS software and a weather generator (Stockle, 1996). Link to economic and risk analysis models is under development. The model's objective is to serve as an analytical tool to study the effect of cropping systems management on crop productivity and the environment. For this purpose, CropSyst simulates the soil water budget, soil-plant nitrogen budget, crop phenology, crop canopy and root growth, biomass production, crop yield, residue production and decomposition, soil erosion by water, and pesticide fate. These are affected by weather, soil characteristics, crop characteristics, and cropping system management options including crop rotation, cultivar selection, irrigation, nitrogen fertilization, pesticide applications, soil and irrigation water salinity, tillage operations, and residue management.

The model code is written in Turbo Pascal (both DOS and Windows versions). An advanced user-friendly interface allows users to easily manipulate input files, verify input parameters for range errors and cross compatibility, create simulations, execute single and batch run simulations, customize utputs, produce text and graphical reports, link to spreadsheet programs, and even select a preferred language for the interface text. Simulations can be customized to invoke only those modules of interest for a particular application (e.g., erosion and nitrogen simulation can be disabled if not desired), producing more efficient runs and simplifying model parameterization. The model is fully documented (Stockle and Nelson, 1994, Stockle and Nelson, 1996), and the manual is also available as a help utility from the CropSyst interface. CropSyst executable program, manual, and tutorials can be retrieved directly over the Internet (http://www.cahe.wsu.edu/~bsyse/faculty/stockle/cropsyst/cropsyst.html).

Although the parameter e (Eq. 2) includes the effect of the temperature regime prevailing during its experimental determination, temperature limitations during early growth are not captured and a single value is determined for the vegetative period or, more usually, for the entire growing season. However, more detailed measurements will show a decrease of e during early growth due to low temperature. Not accounting for this temperature effect may result in overprediction of biomass production during early growth, particularly in the case of winter crops. A temperature limitation factor is included in CropSyst to correct the value of e during this period, which is assumed to increase linearly from zero to one as air temperature fluctuates from the base temperature for development to an optimum temperature for early growth.

To account for nitrogen effects on biomass production, the minimum of B_T and B_L is used as base to determine the nitrogen-dependent biomass production (B_N):

The increase of leaf area during the vegetative period, expressed as leaf area per unit soil area (leaf area index, LAI), is calculated as a function of biomass accumulation, specific leaf area, and a partitioning coefficient. Leaf area duration, specified in terms of thermal time and modulated by water stress, determines canopy senescence. Root growth is synchronized with canopy growth, and root density by soil layer is a function of root depth penetration. The prediction of yield is based on the determination of a harvest index (grain yield/aboveground biomass). Although an approach based on the prediction of yield components could be used, the harvest index seems more conservative and reliable for a generic crop simulator. The harvest index is determined using as base the unstressed harvest index, a required crop input parameter, modified according to crop stress (water and nitrogen) intensity and sensitivity during flowering and grain filling.

Model Inputs
Four input data files are required to run CropSyst: Location, Soil, Crop, and Management files. Separation of files allows for an easier link of CropSyst simulations with GIS software. A Simulation Control file combines the input files as desired to produce specific simulation runs. In addition, the Control file determines the start and ending day for the simulation, define the crop rotations to be simulated, and set the values of all parameters requiring initialization. Definitions, usage, and range of variation of all parameters required by CropSyst are given in the User's Manual (Stockle and Nelson, 1994 and 1996), and they are also available in the Help facility of the model interface.

The Location file includes information such as latitude, weather file code name and directories, rainfall intensity parameters (for erosion prediction), freezing climate parameters (for locations where soil might freeze), and local parameters to generate daily solar radiation and vapor pressure deficit values.

The Soil file includes surface soil Cation Exchange Capacity and pH, required for ammonia volatilization, parameters for the curve number approach (runoff calculation), surface soil texture (for erosion calculation), and five parameters specified by soil layer: Layer thickness, Field Capacity, Permanent Wilting Point, Bulk Density, and Bypass Coefficient. The latter is an empirical parameter to add dispersion to solute transport, particularly when using the cascading approach for soil water redistribution.

The Management file includes automatic and scheduled management events. Automatic events (irrigation and nitrogen fertilization) are generally specified to provide optimum management for maximum growth, although irrigation can also be set for deficit irrigation. Management events can be scheduled using actual date, relative date (relative to year of planting), or using synchronization with phenological events (e.g., number of days after flowering). Scheduled events include irrigation (application date, amount, chemical or salinity content), nitrogen fertilization (application date, amount, source- organic and inorganic-, and application mode- broadcast, incorporated, injected), tillage operations (primary and secondary tillage operations, which are basically related to residue fate), and residue management (grazing, burning, chopping, etc.).

The Crop file allows users to select parameters to represent different crops and crop cultivars using a common set of parameters. This file is structured in the following sections: Phenology (thermal time requirements to reach specific growth stages, modulated by photoperiod and vernalization requirements if needed), Morphology (Maximum LAI, root depth, specific leaf area and other parameters defining canopy and root characteristics), Growth (transpiration-use efficiency normalized by VPD, light-use efficiency, stress response parameters, etc.), Residue (decomposition and shading parameters for crop residues), Nitrogen Parameters (defining crop N demand and root uptake), Harvest Index (unstressed harvest index and stress sensitivity parameters), and Salinity Tolerance.

Table 1. Summary of statistical results for comparisons of simulated and observed yields (from Pala et al., 1996, and Stockle et al., 1994)

d = Willmott Index of Agreement (Willmott, 1982), ranging from 0 to 1, 1 being perfect agreement, B = Biomass, G = Grain Yield, W = Water treatments were imposed, N = Nitrogen treatments were imposed

Recent validation work was performed using data collected by the Institut National de la Recherche Agronomique (INRA) at Auzeville (near Toulouse), France (Stockle et al., 1996). These data are from long-term cropping system experiments conducted from 1983 to 1992 to evaluate crop rotations at three input levels. Input level I was unirrigated and received a minimum amount of fertilization; level II received limited irrigation, restricted to the most sensitive growth phases, and a moderate amount of fertilization; and level III received full irrigation and a large amount of fertilization. The objective was to evaluate the ability of CropSyst to predict ET, biomass, and yield of maize, sorghum, and soybean in response to weather (three dry years: 1986, 1989, and 1990) and soil water availability. In addition, simulations were performed using four combinations of two ET and two infiltration/redistribution submodels. The ET submodels corresponded to the Penman-Monteith (P-M) and Priestley-Taylor (P-T) equations, the latter applied with a VPD-dependent P-T coefficient. Infiltration/redistribution submodels corresponded to the cascading [C] method and the finite difference (FD) method. CropSyst was found able to simulate well the observed ET, biomass, and grain yield for the three crops, three years, and three irrigation input levels as given by Wilmott index of agreement consistently over 0.95. Results in Table 2, which include only crop yield simulations, show that the best simulations tended to be associated with the use of the P-M ET and the FD water transport submodels. However, results using the simpler methods are not too different, which is encouraging for applications where data input or computer CPU time constraints may be an issue.

Table 2. Summary of statistical results for comparisons of simulated and observed grain yield at Auzeville, France using different ET and water transport submodels (PM = Penman-Monteith ET submodel; PT = Priestley-Taylor submodel; C = cascading infiltration; FD = finite difference infiltration)

The capability of CropSyst to simulate different cropping systems using 6 years of data collected from rotation experiments at two locations, representative of the two largest plain areas of Italy, was tested in Donatelli et al (1996a). Simulations were performed by initializing state variables at the beginning of 6-year rotations without further reinitialization, thus constituting a severe test of the model's medium-term predictive capabilities. Data available did not allow for detailed corroboration of model components and limited further analysis for correction of situations where model performance was poor.
Model estimates of yield of maize, soybean, and barley at Modena, and sorghum and sunflower at Foggia, appeared reasonably accurate. CropSyst was not able to simulate soybean growth when the crop was sown as a second crop after durum wheat at Foggia. However, poor simulation of winter cereal yields proved to be the most critical limitation of the model, particularly at Foggia, and the variability observed in durum wheat yields at this location in different rotations could not be explained satisfactorily.

Table 3. Statistical indices to evaluate simulation results at Modena and Foggia, Italy. Key:
n, number of observations;
, average measured yield;
, average simulated yield;
RMSE, root mean square error^†
EF, modelling efficiency^†;
CRM, residual mass coefficient^†;
slope, intercept and r² of the regression predicted vs. measured yield.

	n	t ha-1	t ha-1	RMSE %	EF	CRM	Slope	Int.	r2
MODENA
barley	48	6.01	5.91	7.42	0.59	0.0163	0.53	2.74	0.62
maize	39	9.35	9.44	4.84	0.64	-0.0103	0.81	1.90	0.68
soybean	50	2.90	2.85	13.73	0.85	0.0164	1.06	-0.21	0.89
FOGGIA
durum wheat	70	2.58	2.50	15.59	-0.38	0.0303	0.06	2.34	0.01
sorghum	29	6.53	6.67	19.93	0.57	-0.0217	0.90	0.79	0.67
soybean(sum. sow.)	30	1.99	1.78	18.84	-0.62	0.1074	0.00	1.76	0.00
sunflower	20	3.23	3.24	20.66	0.63	-0.0020	0.69	0.99	0.63

The model was able to simulate correctly water use by crops in different years, but the rewetting of soil profile during the second part of the year was often overestimated for surface soil layers and underestimated for deeper soil layers, presumably as the consequence of a seasonal preferential water flow due to soil cracking. As an example, the following figure shows simulated and measured soil water content fluctuations for the two-years rotation sunflower-durum wheat at Foggia.

Figure 1. Simulated and measured soil water content for the rotation sunflower-durum wheat at Foggia, Italy. Average values of the soil layer .05-.5 m.

Work under progress is applying CropSyst to study the economic risk of selected crop rotations in the Palouse region of the Pacific Northwest, USA. This is a dryland region characterized by steep gradients of precipitation fluctuating from 200 to 500 mm., with weather conditions ranging from excellent to marginal for small grain production. Crop rotations evaluated include Winter Wheat/Spring Barley/Spring Peas, Winter Wheat/ Spring Peas, Winter Wheat/Spring Barley/Fallow, Winter Wheat/Fallow, and continuous Spring Barley. Thirty-year average yield of the different crops within typical rotations have been compared with long-term farm-level yield averages. Both the simulated average and the coefficient of variation for the three crops compared well with observed values. Comparisons for winter wheat and spring barley are shown across the rainfall gradient (Fig. 2).

Figure 2. Simulated and observed long-term yields for winter wheat and spring barley in typical rotations at the Palouse region of the Pacific Northwest, USA (S = Simulated, O = Observed).

Plans for Development
CropSyst improvement is an ongoing and challenging process. In general, the introduction of new management capabilities or new simulation modules is not very likely in the near future, but rather improvement of process simulation will be given priority. The capability of accounting for tillage effects on both infiltration and evaporation will be implemented in the model, and the evaporation process will be re-evaluated to more accurately simulate evaporation under fallow conditions.

Validation with data sets from all over the world is of great interest to ensure robustness of the model. Test of the model with new crops such as potato (in progress), sugarbeet, alfalfa, canola, and others will be attempted as proper data sets become available. Cooperation with agronomists and agricultural scientists around the world is desirable for further progress.

Acknowledgments

We gratefully acknowledge the contribution of P. Debaeke, M. Cabelguenne, E. Ceotto, P. Spallacci, D. Ventrella, and M. Rinaldi

References
Donatelli, M., C.O. Stockle, E. Ceotto, and M. Rinaldi. 1996a. CropSyst validation for cropping systems at two locations of Northern and Southern Italy. European Journal of Agronomy (in press).

Donatelli, M., P. Spallacci, R. Marchetti, and R. Papini. 1996b. Evaluation of CropSyst simulations of growth of maize and of water balance and soil nitrate content following organic and mineral fertilization . Proc. 4th European Society of Agronomy Congress, Veldhoven, The Netherlands (in press)

Ferrer-Alegre, F. 1995. A model for assessing crop response and water management in saline conditions. MS Thesis, Washington State University, Pullman, WA, USA.

Loague, K. and R.E. Green. 1991. Statistical and graphical methods for evaluating solute transport models: overview and application. J. Contam. Hydrol., 7:51-73.

Monteith, J.L. 1977. Climate and crop efficiency of crop production in Britain. Phil. Trans. Res. Soc. London Ser. B, 281:277-329.

Pala, M., C.O. Stockle, and H.C. Harris. 1996. Simulation of durum wheat (triticum durum) growth under differential water and nitrogen regimes in a mediterranean type of environment using CropSyst. Agricultural Systems (in press).

Stockle, C.O. 1996. GIS and simulation technologies for assessing cropping systems management in dry environments. American Journal of Alternative Agriculture (in press).

Stockle, C.O. and R. Nelson. 1996. Cropsyst User's manual (Version 2.0). Biological Systems Engineering Dept., Washington State University, Pullman, WA, USA (In preparation).

Stockle, C.O., M. Cabelguenne, and P. Debaeke. 1996. Validation of CropSyst for water management at a site in southwestern France. Proc. 4th European Society of Agronomy Congress, Veldhoven, The Netherlands (In press)

Stockle, C.O. and P. Debaeke. 1996. Modeling crop N requirement: A critical analysis. Proc. 4th European Society of Agronomy Congress, Veldhoven, The Netherlands (In press)

Stockle, C. O., S. Martin and G. S. Campbell. 1994. CropSyst, a cropping systems model: water/nitrogen budgets and crop yield. Agricultural Systems 46:335-359.

Stockle, C.O. and R. Nelson. 1994. Cropsyst User's manual (Version 1.0). Biological Systems Engineering Dept., Washington State University, Pullman, WA, USA.

Tanner, C.B. and T.R. Sinclair. 1983. Efficient water use in crop production: Research or Research? In Limitations to Efficient Water Use in Crop Production. H.M. Taylor, W.R. Jordan and T.R Sinclair (eds.). Amer. Soc. Agron, Madison, WI, USA.

Willmott, C.J. 1982. Some comments on the evaluation of model performance. Bull. Amer. Meteor. Soc. 63:1309-1313.