Modelling of Chlorophyll-A Concentrations in Pulicat Lagoon , Southeast Coast of India Using Artificial Neural Network (original) (raw)

Owing to its negative impacts on human health and aquatic life, this widely reported phenomenon has become a serious environmental problem. While many process-based, statistical and empirical models exist for water quality prediction, Artificial Neural Network (ANN) models are increasingly being used for water related applications as they are often capable of modelling complex systems for which behavioural rules or underlying physical processes are either unknown or difficult to simulate. In the present study, a feed forward neural network is proposed to model the primary productivity of Pulicat lagoon. The commonly used back propagation learning algorithm has been employed for training the ANN. The model was constructed using five years of seasonal data set on the mouth part of Pulicat lagoon which is the most dynamic part of the lagoon. Despite the very complex and peculiar nature of this region of the lagoon, a very good correlation (R = 0.998) was observed between the measured and predicted values during model validation. The Mean Square Error between the measured and predicted values was found to be 0.018. Thus, the resulting prediction of Chlorophyll-a values clearly indicated that ANNs can fit the complexity and nonlinearity of ecological phenomena such as phytoplankton production to a high degree.