Risk assessment in decision-making using the COTONS model

Off-station experiments carried out in the framework of adaptive research and development programs, are generally limited by technical, human and/or financial considerations. The number of experimental sites needed to represent the environmental conditions variability of the cotton production areas, can be prohibitive. The major experimental designs carried out for adaptive research purpose, are usually not relevant for risk assessment associated with research and out-reach crop management recommendations. The cotton plant model COTONS® allows researchers to assess yield response to various crop management sequences and environmental conditions. These conditions include soil hydraulic characteristics and daily climatic data (solar radiation, rainfall, minimum and maximum temperatures and wind speed). This environmental data is complemented by soil fertility and cultural practices applied to the crop (variety, plant population, fertilisation, plant protection etc...). This paper presents a probability-based approach to assess the risk-associated in decision-making under the various environmental uncertainty using COTONS®. Using this approach, model users are better informed of the risk and cost related to their decision making such as when and how much to apply fertilizer and/or pesticides. According to the proposed approach, daily decision rules for better crop productivity are based on crop state variables, i.e. cultivation techniques, application of additional N, and plant protection, etc. These rules resulted from a simulated data set corresponding to 396 environmental conditions and four crop management sequences. The risk assessment for a specific yield target is calculated as one minus the probability of exceedance the desired yield target level. The output of the risk-based approach for selected decision rules and crop management sequences are summarized in probability of exceedence plots. The plots allow the user the choice of the level of risk they choose to use and the probability trends for each decision rule. An economic cost evaluation of the different strategic options in crop management is also presented. The above approach allows for the assessment of the risk for different cotton crop management sequences. Application of the proposed approach may reduce the risk of obtaining the lowest yields for a certain condition. The COTONS® plant model was demonstrated as a powerful research tool for improved and more informed decision-making. (Résumé d'auteur)