Stream water sampling during hydrological events is important, but also challenging, because stream water chemistry can vary largely during events and the number of samples which can be taken is limited due to practical constrains. This project will contribute towards better observations of the temporal variability of stream water chemistry (including isotope composition) by intelligent sampling strategies. This will support both hydrological research and monitoring. The objectives of this project are (i) the identification of optimal sampling distribution during events based on perfect information (i.e., in hindsight) to maximize the information gained by a limited number of water samples, (ii) the formulation of a sampling strategy under real conditions, where sampling decisions have to be made based on (uncertain) forecasts and (iii) the development of a technical implementation for the instrumentation of a forecast-based sampling based on low-power wireless sensor networks.