Model-based validation of observational data: The OSPAR North Sea eutrophication assessment as a test case

Abstract

Monitoring of marine ecosystems is essential for assessing environmental conditions and informing policy decisions. Observational data, while critical, are often spatially and temporally limited, and improper extrapolation might lead to potential biases. Therefore, in situ observations should be validated before use. This need for validation is demonstrated in the present work, where a Marine Mechanistic Ecosystem Model is used to validate the representativeness of the OSPAR North Sea COMP4 winter nutrient observational data set used for the assessment of eutrophication state. By comparing in situ measurements with high-resolution data generated by the model, we assess the confidence in the observational data and explore the sensitivity of the resulting classification of the nutrient status to different sampling strategies. The results highlight the need for proper validation of the observational data, as the confidence of the final assessment strongly depends on the sampling model. Furthermore, sensitivity analysis reveals that the timing of sampling can substantially impact the eutrophication status assessment, and that the current monitoring strategies may not fully capture ecosystem dynamics, highlighting the need for optimized sampling models. Given the increasing budget constraints on monitoring programs, the combined use of ecosystem models and in situ observations can improve the reliability of marine environmental assessments.