Font Size: a A A

An Integrated Data-Driven Method for the Reservoir Eutrophication Management

Posted on:2017-05-10Degree:Ph.DType:Dissertation
University:North Carolina Agricultural and Technical State UniversityCandidate:Tao, XiaojueFull Text:PDF
GTID:1451390008979832Subject:Environmental Engineering
Abstract/Summary:
Water system eutrophication is a world-wide issue for sustainable development. The aging process of natural water systems has been greatly accelerated by human activities. This phenomenon raises challenges in the behavior prediction of nutrient-enriched water bodies and increases uncertainties in water systems. To address these challenges, an integrated data-driven method was proposed for the reservoir eutrophication management in this dissertation research. This data-driven method integrates trophic state index modification, artificial neural network (ANN) modeling and remediation planning. Since Carlson trophic state index (CTSI) is highly regional-dependent, a modified trophic state index (MTSI) was developed by re-evaluating the pairwise linear relationships among Secchi Disk Depth (SD), Chlorophyll-a (Chl-a) and Total Phosphorus (TP). The case study results demonstrated that the MTSI can evaluate the eutrophication level more accurately than CTSI in the study reservoir. An ANN modeling procedure was designed to model the eutrophication process and predict limiting factors for eutrophication. Following this procedure, an ANN model was developed to predict TP, which is the limiting factor for eutrophication in the study reservoir. The case study results showed that the coefficient of multiple determination (R2) of the ANN model is 0.8575, and the mean squared error of TP prediction is 1.5671x10-4 (mg/L)2 . Finally, the land use management problem in the reservoir watershed was formulated as a two-objective optimization model. Then a general simulated annealing algorithm was extended to discover the Pareto front of the two-objective optimization problem and generate optimal land use plans for stakeholders to choose according to their preferences. The case study results demonstrated that the integrated data-driven method proposed can accurately predict the limiting factor(s) for eutrophication in a water system, and can recommend the best land use plans of a reservoir watershed to prevent eutrophication.
Keywords/Search Tags:Eutrophication, Reservoir, Integrated data-driven method, Water, Case study results, ANN
Related items