A comparative study of hydrodynamic model and expert system related models for prediction of total suspended solids concentrations in Apalachicola Bay

SummaryThis paper presents the development of expert system related models using the concepts of Geno-Kalman Filtering (GKF) and fuzzy logic to predict the concentration of total suspended solids (TSS) with the input of wind speed at the locations of Cat Point and Dry Bar in Apalachicola Bay, USA. The TSS and wind speed data recorded from June 1, 2005 to July 30, 2005 are used for the present modeling study. Data set is divided into two parts for the model development. The June data are selected for calibration and the July data are for model verification. The predicted TSS concentrations from the Geno-Kalman Filtering and fuzzy logic models are compared with those from a hydrodynamic model and measured data reported in Liu and Huang (2009). The statistical values in terms of the mean squared error (MSE), the coefficient of efficiency (CE) and the chi-square (χ2) parameter between the observed data and predicted results for the evaluation of each model’s performance are presented. It is noted that both the hydrodynamic and Geno-Kalman Filtering models are capable of predicting TSS concentration with the trend of dynamic variation. Although the fuzzy logic and the Geno-Kalman Filtering models outperform the hydrodynamic model, the Geno-Kalman Filtering model is found to be able to produce the most accurate results at the studied locations. The fuzzy logic model tends to under-predict the TSS concentrations.

Research highlights
► Geno-Kalman Filtering and fuzzy logic models are developed to predict the total suspended solids concentrations.
► Both the Geno-Kalman Filtering and fuzzy approaches outperform the hydrodynamic model.
► The Geno-Kalman Filtering performs better than fuzzy modeling.
► The fuzzy logic model is practical for the prediction applications.