Modeling algal atypical proliferation using the hybrid DE–MARS–based approach and M5 model tree in La Barca reservoir: A case study in northern Spain

• Published in: Ecological engineering Vol. 130; pp. 198 - 212
• Main Authors: García-Nieto, P.J., García-Gonzalo, E., Alonso Fernández, J.R., Díaz Muñiz, C.
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.05.2019; Elsevier BV
• Subjects: Adaptive algorithms; Algae; Algal abnormal productivity in reservoirs; Aquatic ecosystems; Biological fertilization; Case studies; Coefficients; Differential evolution (DE); Ecosystems; Environmental degradation; Eutrophication; Evolutionary computation; Fertilization; Lakes; M5 model tree; Modelling; Multivariate adaptive regression splines (MARS); Optimization; Organic chemistry; Proliferation; Regression analysis; Reservoirs; Restoration; Splines; Training; Water management; M5 model tree; Multivariate adaptive regression splines (MARS); Regression analysis; Algal abnormal productivity in reservoirs; Differential evolution (DE)
• ISSN: 0925-8574; 1872-6992
• DOI: 10.1016/j.ecoleng.2019.02.020

•A hybrid DE–MARS–based model is a good predictive model of the phosphorus and chlorophyll presences.•Algal abnormal proliferation is dangerous for environment in lakes.•The biological and physical-chemical input variables in this process are studied in depth.•The results of this hybrid model for the eutrophication evaluation is compared with the M5 model tree.•The obtained correlation coefficients of two hybrid DE–MARS–based models are about 90%. Algal atypical proliferation is a consequence of water fertilization (also called eutrophication) and one of the main causes of the degradation of reservoir and lake ecosystems. Its intensification during the last decades has led the stakeholders to seek water management and restoration solutions, including those based on modelling approaches. In this way, this paper presents one reservoir eutrophication modelling based on a new hybrid algorithm that combines multivariate adaptive regression splines (MARS) and differential evolution (DE) to estimate the algal abnormal proliferation from physical-chemical and biological variables. This technique involves the optimization of the MARS hyperparameters during the training process. Additionally, an M5 model tree was fitted to the experimental data for comparison purposes. Apart from successfully forecasting algal atypical growth (coefficients of determination equal to 0.83 and 0.91), the model showed here can establish the significance of each biological and physical-chemical parameter of the algal enhanced growth. Finally, the main conclusions of this research work are exposed.