Assembly of a model for grapevine powdery mildew in a decision support system and search for evaluation criteria

Decision making in vine protection against powdery mildew is complex since this disease must be treated before symptoms appear. To trigger treatments only when needed, decision support systems should combine precise models with decision rules. The latter may have to be adapted according to climate and national policies. Therefore, model simulated variables must be rich enough to feed different rules. This work describes a methodology to design such a model from heterogeneous scientific literature. It also reviews to what extent observation of severity databases enables evaluation of the quality of model and decision rules.