Abstract Background Water quality assessment must rely on representative samples, but obtaining them is increasingly challenging given the growing number and diversity of trace contaminants emitted into water bodies. This study investigates whether and to what extent monthly grab sampling (GS), widely used in official monitoring programmes, provides reliable and accurate estimates of annual mean and maximum concentrations, and of annual riverine loads. Through a one-year systematic survey in two rivers using three sampling techniques in parallel, we evaluate low-frequency GS against time-proportional composite (TPC) and flow-proportional composite (FPC) sampling. The scope of 450 contaminants encompasses potentially toxic elements (PTE), pharmaceuticals, biocides, pesticides and PFAS, covering a broad diversity of emission patterns and environmental fate. Results For dissolved PTEs and pharmaceuticals, monthly GS delivered approximately correct annual average concentrations, but potential underestimation and overestimation were also observed. Contaminants with seasonal patterns or those emitted during short-term events, such as many pesticides and biocides, PFAS and total PTEs, were not adequately depicted by low-frequency GS, leading to varying degrees of underestimation and seldom also to overestimation. This applies to average annual concentrations but particularly to loads. Integrated composite sampling, especially flow-proportional, performs significantly better for such contaminants. Worthy of attention are cases of low and highly variable concentrations, such as ibuprofen, lindane and PFNA in this study. While integrated composite samples clearly demonstrated their non-negligible presence, monthly GS failed to detect them. Conclusion The widely applied monitoring approach relying on monthly GS is only sufficiently reliable for contaminants being emitted fairly constantly and transported primarily in the dissolved phase. Even then, its good performance is mostly limited to assessing average concentrations and thus chronic exposure, while integrated composite sampling can significantly improve the accuracy of load calculations. For contaminants with variable and dynamic concentration patterns, FPC sampling performs considerably better, especially in terms of load calculations, followed by TPC sampling. However, the latter is simpler to implement. GS-based underestimations pose a significant risk of incorrectly assessing compliance with environmental quality standards if the concentration level in rivers is not very far below or above the thresholds, as was often the case in our study area.
Weber et al. (Mon,) studied this question.