Forecasting red tides using an hybrid neuro-symbolic system

Spain. system which possesses the flexibility to behave in different ways depending on the state of the environment. This paper presents the application of a novel hybrid artificial intelligence (AI) model to a forecasting problem over a complex and dynamic environment. The approach, which is discussed, is capable of producing satisfactory results in situations in which neither artificial neural network nor statistical models have been sufficiently successful. The oceans of the world form a highly dynamic system for which it is difficult to create mathematical models [1]. Red tides are the name for the discolourations caused by dense concentrations of microscopic sea plants, known as phytoplankton. The discolouration varies with the species of phytoplankton, its pigments, size and concentration, the time of day, the angle of the sun and other factors. Red tides usually occur along the north west coast of the Iberian Peninsula in late summer and autumn [2]. The prevailing southerly winds cause cold, nutrient-rich water to rise up from the deeper regions of the ocean to the surface, a process known as upwelling. Swept along with this upwelled water are dinoflagellate cysts, the resting stages of the organism, which lie dormant in the sediments on the sea floor. The high nutrient concentrations in the upwelled water, together with ideal conditions of temperature, salinity and light, trigger the germination of the cysts, so that the dinoflagellates begin to grow and divide. The rapid increase in dinoflagellate numbers, sometimes to millions of cells per liter of water, is described as a bloom of phytoplankton (concentration levels above the 100.000 cells per liter). Concentration of the bloom by wind and currents, as well as the dinoflagellates’ ability to swim to the surface, combine to form a red tide. If conditions at the water’s surface become unfavourable for the dinoflagellates, for example, if the nutrients are depleted or the bloom is dispersed by wind and currents, the dinoflagellates will again form dormant cysts and sink to the sea floor. This study focusses on the