An ocean circulation forecasting model for the Madeira Archipelago is operational since May 2010. Developing a forecasting system for a small island oceanic region, deprived from in-situ observations, is a challenging task since there are limited ways to validate predictions. Furthermore, model resolution concurrent with insufficient computational power, locally available, are other limiting factors to consider. Regional models combined with the possibility to downscale solutions onto a higher resolution island-scale model is a way to overcome some of such limitations. Nevertheless, generalised regional models must be able to accurately represent the far-field and transport important features such as meddies onto the local systems; while island-scale models must have sufficient grid resolution as well as adequate physics and accurate atmospheric forcing to resolve the near-field phenomena. An island-induced cyclonic eddy event was successfully observed and forecasted with the current approach (regional-local). Generalised single (regional) model initiatives will prove to be insufficient to deal with mesoscale dynamic systems, islands and seamounts are important generators of mesoscale features in the NE Atlantic, with basin scale implications. The forecasting systems of the future should also consider upscaling valid local (island-scale) solutions onto Regional and/or Global models.

The Island Mass Effect has been primarily attributed to nutrient enhancement of waters surrounding oceanic islands due to physical processes, whereas the role of land runoff has seldom been considered. Land runoff can be particularly relevant in mountainous islands, highly susceptible to torrential rainfall that rapidly leads to flash floods. Madeira Island, located in the Northeast Atlantic Ocean, is historically known for its flash flood events, when steep streams transport high volumes of water and terrigenous material downstream. A 22-year analysis of satellite data revealed that a recent catastrophic flash flood (20 February 2010) was responsible for the most significant concentration of non-algal Suspended Particulate Matter (SPM) and Chlorophyll-a at the coast. In this context, our study aims to understand the impact of the February 2010 flash flood events on coastal waters, by assessing the impact of spatial and temporal variability of wind, precipitation, and river discharges. Two specific flash floods events are investigated in detail (2 and 20 February 2010), which coincided with northeasterly and southwesterly winds, respectively. Given the lack of in situ data documenting these events, a coupled air-sea-land numerical framework was used, including hydrological modeling. The dynamics of the modeled river plumes induced by flash floods were strongly influenced by the wind regimes subsequently affecting coastal circulation, which may help to explain the differences between observed SPM and Chlorophyll-a distributions. Model simulations showed that during northeasterly winds, coastal confinement of the buoyant river plume persisted on the island’s north coast, preventing offshore transport of SPM. This mechanism may have contributed to favorable conditions for phytoplankton growth, as captured by satellite-derived Chlorophyll-a in the northeastern coastal waters. On the island’s south coast, strong ocean currents generated in the eastern island flank promoted strong vertical shear, contributing to vertical mixing. During southwesterly winds, coastal confinement of the plume with strong vertical density gradient was observed on the south side. The switch to eastward winds spread the south river plume offshore, forming a filament of high Chlorophyll-a extending 70 km offshore. Our framework demonstrates a novel methodology to investigate ocean productivity around remote islands with sparse or absent field observations.

No abstract