SeaWIFS Coastal Waters Mapping Using an Adaptive Neuro-fuzzy Inference System (ANFIS)

Monitoring coastal water quality is an extremely important factor as it affects the ecological balance of ecosystems and the development of the social and economic wellbeing of the countries bordering it. Actually, remote sensing data enables us to monitor number of water-quality indicators that reflect the state of the entire coastal ecosystems. These indicators observable from space help us to estimate among others the concentrations of Chlorophyll (Chl), Suspended Particulate Matter (SPM), as well as water turbidity (Turb). In this paper, we focused our interest to coastal water quality mapping process using the combination of in-situ measurements and SeaWIFS data. Water parameters map’s is processed by Adaptive Neuro-Fuzzy Inference System (ANFIS). Thus, we evaluated the predictive performance of ANFIS with four different membership functions (MFs): triangular, trapezoidal, Gaussian and sigmoid functions. Considering an ANFIS structure of a single output (uncorrelated water parameters), the most accurate map was generated with a Gaussian membership function, yielding a correlation coefficient of 70%.