Robust water quality model calibration using an alternating fitness genetic algorithm

Presented herein is a robust approach to calibrating water quality models for water quality management using sparse field data. The calibration procedure adopts genetic algorithms (GAs) to inversely solve the governing equations, along with an alternating fitness method to maintain solution diversity. The proposed approach is illustrated with a total phosphorus model of the Triadelphia Reservoir in Maryland. A series of deterministic and stochastic alternating fitness GA schemes are implemented and compared with a standard GA. Significantly higher diversity is observed in the solutions obtained by the alternating fitness method than by the standard process. The diversified solutions obtained by the alternating fitness GA method are then classified into several patterns using a parameter pattern recognition model. The best solutions to each pattern are then chosen for further projection analyses, which generate a range of prediction results that provide decision makers with information for formulating sound pollution control schemes.