Prediction model of water quality in Chaophraya River using artificial neural network

Abstract

Water quality prediction models have been researched, developed and applied to various water resources around the world. The characteristics of different water resources make the model to forecast the water quality parameters of Chaophraya River in the future that was developed from artificial neural network (ANN). Unlike the traditional ANN, the proposed model called space and time neural network (STNN) is able to accept input multi-dimensional data (historical and upstream water quality records). The state-of-the-art models and the purposed model were tested and compared using the Chaophraya River’s water quality measured over a period of 17 years. The STNN model outperforms the others in term of water quality prediction correlation coefficient (Spearman’s rho = 0.73 ± 0.06) compared with other models (Spearman’s rho = 0.67 ± 0.08 and (Spearman’s rho = 0.57 ± 0.15). In addition, the proposed model was developed with a general framework and could be applied to other rivers as well.