Solitary Wave Run-Up Over Fringing Reefs And Its Prediction Based On The Artificial Neural Network

Tsunami is a kind of natural disaster mainly caused by submarine earthquake,volcanic eruption,submarine collapse and landslide and other earth activities,which causes submarine water disturbance and generates waves.Tsunami moves forward at a very fast speed and poses a great threat to the infrastructure in coastal areas and the safety of residents' lives and property.Because the leading wave of the tsunami wave is close to the solitary wave,the physical model experiments and numerical simulation experiments of the tsunami wave are usually simplified to the solitary wave.Fringing reef topography,as a typical nearshore topography,is of great practical significance to study the transformation and run-up characteristics of solitary waves on this topography.First,a series of laboratory experiments is carried out in a wave flume to study the influence of rough reef surface on solitary wave transformation and run-up over coral reefs.The cylinder array is used to model the reef surface with large roughness.The experiment mainly uses video capture system combined with wave height acquisition system to test the combination of a series of solitary wave conditions under the smooth and rough reef.Furthermore,based on the Boussinesq-type CoulWave models,a reef numerical model is established.The friction coefficient is used to characterize the reef roughness.The feasibility of the numerical model is verified by experimental data.The influence of two hydrodynamic forcings(incident wave height,reef-flat water level)and four reef morphologic features(reef width,fore-reef slope,beach slope,reef roughness)on wave run-up on the back-reef beach is discussed by using this model.Finally,two hydrodynamic forcings(incident wave height,reef-flat water level)and four reef morphologic features(reef width,fore-reef slope,beach slope,reef roughness)are selected as the input variables and wave run-up on the back-reef beach is assigned as the output variable.The MLP neural network model is established to predict wave run-up on the back-reef beach and applied to the physical model experiment.The results show that the rough reef surface significantly reduces the leading solitary wave and the secondary wave due to beach-reef beach reflection,as well as the wave speed on the reef flat.The relative cross-shore wave height decreases with the increase of dimensionless incident wave height along the reef.It also increases with the increase of reef-flat wave level.The cross-shore wave height attenuation is more evident with the rough reef surface.The variation of dimensionless reflected wave height with the dimensionless incident wave height depends on the reef-flat wave level.The dimensionless reflected wave height approaches to a constant when the incident wave height is sufficient large.The rough reef surface slightly enhances the wave reflection from the fore-reef slope.The dimensionless transmitted wave height as well as wave run-up decline with the increasing dimensionless incident wave height,particularly for the large reef-flat water level.The dimensionless wave run-up on the back-reef beach with the rough reef surface reduces by an average of 46%compared to that with the smooth surface.An empirical formula is obtained by a regression analysis to predict the wave run-up with both smooth and rough reefs.In this paper,by using numerical model CoulWave based on the Boussinesq equations,a coral reef numerical model is established.The wave run-up of solitary waves on the back-reef beach is studied.The numerical calculation results are in good agreement with the results of experiments.It is shown that the numerical model can better predict the wave run-up under the rough reef surface,and the influence of hydrodynamic forcings and reef morphologic features on the wave run-up of solitary waves on the back-reef beach is further demonstrated.The results show that,the relative cross-shore wave height decreases with the increase of incident wave height and reef roughness along the reef.It also increases with the increase of reef-flat wave level and fore-reef slope.The relative cross-shore wave height decreases with the increase of reef width around beach slope.It also increases with the increase of beach slope.Wave run-up of solitary waves on the back-reef beach increases with the increase of incident wave height,reef-fat water level,fore-reef slope,beach slope,but decreases with the increase of reef width and the reef roughness.This provides a certain reference value for the protection of the shoreline.A validated numerical model CoulWave based on the Boussinesq equations is applied to provide a dataset consisting of 4096 runs for MLP-NN training and testing.In this study,the statistical indexes of the neural network model were analyzed and optimized,and the precision of model performances in view of individual input variables was compared.Results analyses show that the MLP-NN consisting of one hidden layer with ten hidden neurons provides the best predictions for the wave run-up.Subsequently,model performances in view of individual input variables are accessed via an analysis of the percentage errors of the predictions.Finally,a mean impact value analysis is also conducted to evaluate the relative importance of the input variables to the output variable.In general,the adopted MLP-NN has high predictive capability for wave run-up over the reef-lined coasts,and it is an alternative but more efficient tool for potential use in tsunami early warning system or risk assessment projects.