Author(s): Ekaterini Hadjisolomou; Konstantinos Antoniadis; Ioannis Thasitis; Rana Abu-Alhaija; Herodotos Herodotou; Michalis Michaelides

Linked Author(s): Ekaterini Hadjisolomou

Keywords: Chlorophyll-a; Artificial neural network; Anthropogenic pressure; Management

Abstract: In this study, sea surface chlorophyll-a (Chl-a) levels are modelled with the use of artificial neural networks (ANNs) in an attempt to improve the understanding of marine eutrophication, a major environmental problem of the modern world, and explore remediation scenarios. Typical water quality parameters such as water temperature, nitrogen species, phosphorus, pH, salinity, electric conductivity, and dissolved oxygen served as the model’s input. These parameters and Chl-a were measured for several locations near the coastline of Cyprus (located in Eastern Mediterranean) during the period 2000-2014 and were used to build an ANN model. Generally, the marine water quality of Cyprus is good and characterized as ultra-oligotrophic, with very few recorded eutrophication events. However, some of the monitoring stations have been defined in suitable sites for the monitoring of anthropogenic pressure (e.g., due to aquaculture or nearby industrial units). Additionally, the Eastern Mediterranean area is one of the most affected areas by climate change and the associated water temperature increase. The ANN managed to predict with good accuracy the Chl-a levels (r=0.87 for the test set). Sensitivity analysis was also performed, where the input parameters were perturbated for a small change (8% in our case) and the impact of these perturbations was simulated. Based on these simulations, several useful results were derived. The impact of global warming on eutrophication is worrisome since the water temperature increase resulted in an increase of Chl-a levels by over 200%. Analogous behavior was observed for the nutrients’ perturbations, where a huge increase of Chl-a was calculated. These findings indicate that the Cyprus marine environment is fragile and prone to eutrophication under anthropogenic pressure and climate change. The constructed ANN model can serve as a management tool, based on which nutrients threshold values can be calculated in order to maintain the good environmental status of Cyprus coastal water

DOI: https://doi.org/10.3850/IAHR-39WC2521711920221724

Year: 2022