Sampling strategies based on sets of behavioural parameters. Application to a simple pesticide fate model

Sampling strategies are necessary to minimise the cost of data acquisition. We consider the case when several samples must be analysed at the same time, knowing rainfall data, optionally discharge data. We assume also that a rainfall-runoff and pesticide transfer model is available. Three methods are presented based on behavioural sets calculated on prior events, which provide optimal sampling times with respect to different statistical criteria, for a fixed number of samples. A methodology to compare these strategies, adapted to a context of data scarcity is proposed for proper validation. We consider that the combination of a model and a strategy is efficient if it does better than a strategy without a model based on equal volumes. Even with a simple and relatively poor model, it can be concluded that the different strategies are efficient and show similar efficiency in spite of their different levels of conceptual and algorithmic complexity.