Statistical Detection of Spatio-Temporal Patterns in the Salinity Field Within an Inter-Tidal Basin

Salinity is a key factor affecting biological processes and biodiversity in estuarine systems. This study investigates temporal and spatial changes in salinity at a basin-wide scale for 2005–2015 in the Dutch Wadden Sea. Scan statistics is applied to track salinity variations systematically and to detect potential clusters, i.e., estuarine regions marked by anomalous high-salinity (or low-salinity) values in a certain period (i.e., strong deviations from the expected value in a statistical sense). Clusters’ statistical significance has been assessed via Monte Carlo simulations. Particular attention is devoted to event-driven spatial and temporal patterns characterized by extreme salinity values since these episodes dramatically increase stress levels on organisms living in intertidal areas. Periodic components in the modeled salinity time series are identified using wavelet analysis and eventually removed from the signal before performing scan statistics. Wavelet analysis suggests that tides are the chief agent controlling salinity fluctuations in the system at within-day time scales, whereas no dominant periodicities were detected at longer time scales. Scan statistics reveal long-lasting clusters next to the main freshwater outlets and within the areas characterized by low water exchanges. In contrast, active regions of the estuary can efficiently counteract extreme events and quickly recover their pre-perturbation conditions. Finally, by analyzing the freshwater dispersal in the system, it is found that clusters’ occurrence is related to episodic events characterized by extreme conditions in the southwesterly wind and freshwater discharge. This research demonstrates that scan statistics can be used as a powerful tool for spatiotemporal analyses of marine systems and for identifying data-clustering that may be indicative of emerging environmental hazards (e.g., due to climate change).