Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 4/15/2011
Publication Date: 4/19/2011
Citation: White, J.W., Thorp, K.R. 2011. Simulating Stochastic Crop Management in Cropping Systems. Meeting Abstract,41st Biological Simulation Conference, April 19 - 21, 2011, The University of Texas, Austin, Texas.
Technical Abstract: Introduction -- Crop simulation models are uniquely suitable for examining long term crop responses to environmental variability due to changes in climate or other factors. Long-term studies typically emphasize variability related to weather conditions; certain weather-dependent cropping practices may be varied (e.g., for planting date or irrigations), but the inherent variability of cropping practices and of the soil environment are seldom considered. This paper examines strategies for increasing the stochasticity of crop management in cropping system simulations. The basic approach is to assume that each management practice has a component of uncertainty or random error. The underlying hypothesis is that including sources of variability will substantially increase the overall variation in yield and other outputs relative to conventional simulations of long term cropping sequences. Approach -- Among factors potentially influencing variability in crop management are weather and soil conditions, labor and machinery availability, the attitude of a given producer toward risk, and accuracy of field operations per se. To partially capture such variability, we developed a prototype database to describe variability in management of cropping systems. This Cropping Systems Scenario database (CSSdb) was populated with information on two example systems based on published descriptions from extension bulletins. Practices that were addressed included cultivar selection, sowing, fertilization, irrigations (if used), and tillage. Much of the CSSD architecture parallels that of the ICASA Standards (Hunt et al., 2001). Impacts of variable management were simulated using the Cropping Systems Model (CSM), as implemented in version 4.5 of the Decision Support System for Agrotechnology Transfer (DSSAT4.5, Hoogenboom et al., 2010). To simulate a 20-year cropping sequence, we stochastically modified values in the control file for CSM. For discrete variables such as cultivars, selections were based on probabilities based on reported frequencies (e.g., as derived from reports on areas sown to different cultivars). For continuous variables, simple probability density functions were used. For each variable of interest, values were selected randomly based on the expected distribution, and the values used to create a control file representing a cropping sequence where management varied for each crop and each planting. At this time, no interactions among variables are considered. Thus, a late sowing date would not be associated with selection of an earlier-maturing cultivar or reduced fertilizer rate. For testing purposes, two hypothetical scenarios were considered: a high-input, irrigated wheat crop in Arizona, and a low-input, rainfed spring wheat crop in North Dakota. Control files were created with a Python script, which also launches CSM and partially processes the model outputs. Initial Test Results -- Three major challenges have arisen in this preliminary effort. The foremost is that published descriptions of cropping systems seldom include enough information to accurately characterize the prototype CSSdb. The second is that a mechanism is needed to represent interacting conditional responses of farmers, such as choices of cultivars or crops based on performance of previous crops, recent weather, available soil moisture or other factors. The third is that because the architecture of the CSM model does not permit altering input parameters once a simulation run is initiated, values in the control file that logically should be adjusted according to the state of the system cannot be modified. This makes it especially difficult to specify stochastic water and nutrient management that still responds to weather-related variability. References -- Hoogenboom, G., Jones, J. W., Wilkens, P. W., Porter, C. H., Batchelor, W. D., Hunt, L. A., Boote, K. J