Predicting Coastal Dissolved Oxygen Values with the Use of Artificial Neural Networks: A Case Study for Cyprus

Abstract

Coastal hypoxia is a serious environmental problem that needs to be addressed at a global level. The phenomenon of hypoxia is characterized by low Dissolved Oxygen (DO) levels in the water column that causes detrimental effects on aquatic organisms. Anthropogenic activities such as intensive agriculture practices and point-source nutrient loading are considered the main causes of coastal hypoxia. This study utilizes data-driven modelling based on Artificial Neural Networks (ANNs), and specifically Feed-Forward ANNs, to predict surface DO levels. Several surface water quality parameters such as water temperature, nitrogen species (ammonium, nitrite and nitrate), phosphorus, pH, salinity, electrical conductivity, and chlorophyll-a served as the ANN’s input parameters. These parameters were measured at several sampling sites in the coastal waters of Cyprus and some of the sites were located near areas with anthropogenic activities, during the period 2000-2021. An ANN with a 9-5-1 topology was developed and ANN managed to predict with good accuracy the DO levels, with the Coefficient of determination (r 2) as high as r 2=0.991 for the test set. Additionally, sensitivity analysis was performed to measure the impact of each input parameter on the DO level, and it was estimated that the water temperature is the most influential factor. The “Weights” sensitivity analysis algorithm was used for this purpose. The ANN-based method proposed can be used as a management tool for predicting the DO levels and prevention of hypoxia. Therefore, this work has a positive impact on marine sciences and marine information technology.