| Popis: |
Abstract The present study focuses on the Ría de Vigo (NW Spain), a coastal embayment influenced by the Canary Current Upwelling System, which is among the world’s significant Eastern Boundary Upwelling Ecosystems. The research assesses historical changes in the marine carbonate system by generating 25-year weekly time series at six stations . Assessing ocean acidification in the region is complex due to diverse factors influencing coastal carbon dynamics, making predictions more challenging. To capture the specific dynamics in Ría de Vigo, ensembles of Neural Networks were applied. These networks were trained with a data set obtained in several oceanographic cruises, in order to retrieve pH, hydrogen ion concentration and alkalinity, achieving a root mean square error of 0.0272 pH units, 0.588 nmol $$\hbox {kg}^{-1}$$ kg - 1 , and 10.6 $$\upmu$$ μ mol $$\hbox {kg}^{-1}$$ kg - 1 , respectively. Subsequently, time series of the selected variables were generated, applying data of predictors measured at the aforementioned stations . An increase in normalized alkalinity was observed for all stations, except in the surface layer at the innermost location. A decrease in pH and an increase in hydrogen ion concentration were observed for all points, with trends that exceed reported rates of ocean acidification in the open ocean. |
