No abstract

Eutrophication episodes are common in freshwater and coastal environments, causing significant damage to drinking water and aquaculture. Predictive models are efficient approaches for anticipating eutrophication or algal blooms because ecologists and environmentalists can estimate water pollution levels and take appropriate precautionary steps ahead of time. In aquatic ecosystems, chlorophyll-a (Chl-a) can be employed as a water quality indicator, revealing information on man-made physical, chemical, and biological changes variations or seasonal interventions. In the present study, a Seasonal AutoRegressive Integrated Moving Average (SARIMA) model was developed to forecast monthly Chl-a concentrations in the North Lagoon of Tunis, a Ramsar site, and one of the most important lagoons in Tunisia, using approximately three decades of historical data, starting from January 1989 to April 2018. SARIMA (2,0,2)(2,0,2)12 was found to be the best-fitting model for Chl-a forecasting in the North Lagoon of Tunis. The resulting SARIMA model was validated with actual monthly Chl-a concentrations from our last observations. Furthermore, with only one input variable, the SARIMA model showed greater applicability as a eutrophication early warning system using actual past Chl-a data. Finally, the SARIMA model was utilized to anticipate Chl-a levels from May 2018 to December 2025 as an early warning system for ecosystem managers and decision-makers for next generations.

An Artificial Neural Network (ANN), a Machine Learning (ML) modeling approach is proposed to predict the ecological state of the North Lagoon of Tunis, a shallow restored Mediterranean coastal ecosystem. A Nonlinear Auto Regressive with exogenous input (NARX) neural network model was fitted to predict Chlorophyll- a (Chl- a ) concentrations in the North Lagoon of Tunis as an eutrophication indicator. The modeling is based on approximately three decades of monitoring water quality data (from January 1989 to April 2018) to train, validate and test the NARX model. The most relevant predictor variables used in this model were those having a high permutation importance ranking with Random Forest (RF) model, which simplified the structure of the network, resulting in a more accurate and efficient procedure. Those predictor variables are secchi depth, and dissolved oxygen. Various model scenarios with different NARX architectures were tested for Chl- a prediction. To verify the model performances, the trained models were applied to field monitoring data. Results indicated that the developed NARX model can predict Chl- a concentrations in the North Lagoon of Tunis with high accuracy (R= 0.79; MSE= 0.31). In addition, results showed that the NARX model generally performed better than multivariate linear regression (MVLR). This approach could provide a quick assessment of Chl- a variations for lagoon management and eco-restoration.