Partitioning the ocean using dense time series of Earth Observation data. Regions and natural boundaries in the Western Iberian Peninsula

Biogeographic partitioning systems provide an important framework under which ocean management practices can be designed and implemented. Sea Surface Temperature (SST) provides important insights into the features and processes governing biogeography and the natural boundaries that separate regions (fronts). In this paper, a new methodology is suggested to segment the ocean at a subset of an Eastern Boundary Current system (Western Iberian Peninsula). To this end, a dense time series of daily MODIS data (Sea Surface Temperature (SST) and Chlorophyll-a concentration, 2003-2014) was analyzed using Dynamic Time Warping (DTW). The resulting dissimilarity matrix was then clustered using UPGMA to generate a set of dynamic regions. To determine whether the boundaries separating regions coincided with areas of maximum frontal activity, a matching time series was produced for the study area. The methodology successfully delivered a set of nested regions, explained by known oceanographic features and separated by natural boundaries collocated with frontal activity maxima (ANOVA, p < 0.01). The frontal activity, which was also found to be spatially organized (Moran's I > 0.82 and z-score > 6.95, for a p < 0.01) was thus correlated with the dissimilarity patterns of SST. Together, they offer a new perspective into the biogeographic constraints influencing an important ecosystem. SST-based regions were more informative (k= 4) but both the Chlorophyll-a and the multivariate (SST and Chlorophyll-a) alternatives reproduced the coastal/oceanic dichotomy of established systems in a dynamic way. The method was also applied to yearly subsets, which generally matched the regions delineated with the full time series while depicting inter-annual variability. The DTW and the frontal activity maps contributed to the definition of a scalable and ecologically-meaningful system, which can be adopted for the monitoring of marine areas using different (annual to decadal) time scales and easily accessible datasets.