The effects of nonmigrating mid-bar to lateral migration / Zamsalwani Zamri

Lateral migration is the net outcome of erosion and deposition processes. It initiates and develops width adjustment and planform evolution. Lateral migration process involves adjustment of the planimetric geometry and hydraulic factors that lead to sediment deposition on the river bed. Formation of mid-bar is a result of erosion and deposition process in the river. Many structures built within the vicinity of the river such as bridges, waterways/ navigation and flood control contraptions are affected by lateral migration processes and non-migrating mid-bar as a rigid structural in the middle of the channel. Effects of variation of discharge, the morphology evolution is affected and influences expansion area of mid-bar, low-curvature degree in the channel planform and presence of obstacle. The research on the effects of non-migrating mid-bar was carried out on a physical river channel model measuring 47 m long, 2.4 m wide and 1.8 m deep located at Universiti Teknologi MARA Puncak Alam. The non-migrating mid-bar was built between 8 m to 11 m from the V-notch location. The experiment was designed with various discharge (13.53 m3/hour, 10.91 m3/hour, 8.62 m3/hour and 6.64 m3/hour) and degree of deflection used in the experiment are 37 degree, 25 degree and 15 degree. Bed elevation, bankline point and velocity were measured at specified locations. Correlation analysis between the lateral migration rate, M and mean velocity, Umean to other variables were further examined and analysed for the development of a predictive relationship using discrepancy ratio (DR) and rootmean- square error (RMSE) as validation methods. Statistical analysis, artificial neural network (ANN) and linear least squares (LLS) were conducted using the selected variables obtained through dimensional analysis in order to identify the relationship between lateral migration rate to hydraulic characteristics, water properties, soil capacity and gravitational acceleration variables. Most of these methods of analysis were performed to establish the empirical equation of lateral migration rate as a dependent variable and the other variables as independent variables. The selection of the most appropiate variable is based on the value of r-squared (R2), DR and RSME percentage. The selected predictive variables explain the effect of non-migrating bar to lateral migration. The analysis concluded that ANN predicts lateral migration rate satisfactorily. The percentage of accuracies for this model based on R2, DR and RMSE are 96.40%, 100% and 7.44% respectively.