Hydro KNIME: Scientific workflows for reproducible decision support in identifying suitable hydrological reference station sites

Environmental decision support relies on complex analytical processes, including data integration, spatial analysis, assessment, and validation. In practice, the complexity of these processes, combined with limited transparency and reproducibility, can significantly affect the outcomes of critical environmental decisions, such as selecting sites for new hydrological reference stations. Without transparency and reproducibility, it becomes difficult to trace errors, which can result in suboptimal station placement, inaccurate data collection, poorly defined water levels, and delayed emergency responses to potential flood events. Therefore, improving transparency, documentation, and the reproducibility of analytical processes are essential for effective decision support in hydrology management and planning. This paper presents a scientific workflow approach for reproducible decision support to identify suitable sites for new hydrological reference stations. Specifically, we developed Hydro KNIME (Konstanz Information Miner) scientific workflows that integrate a non-dominated sorting genetic algorithm (NSGA) with a set of evaluation criteria to optimize and identify suitable locations. Using a case study of the Pungwe River Basin in Mozambique, we demonstrate how the proposed scientific workflow approach improves the suitability of the identified sites compared to existing hydrological reference stations in the basin. More importantly, we show that enhancing transparency and provenance tracking in decision-making through scientific workflows can help managers and planners better manage the complexity of analytical processes in environmental decision support, such as identifying suitable sites for hydrological reference networks.