A Simple Approach to Predict the Beach Vulnerability to Storm-induced Erosion in Hasaki Coast, Japan

Abstract

 While coastal areas hold various social and economic activities, they are highly vulnerable to frequent hazards that cause instabilities to community lives. An index-based approach, established on set criteria, is commonly used to quantify and analyze the vulnerability towards specific hazards. This study aims to assess the performance of a simple linear regression model in predicting storm-induced erosion and to propose a vulnerability assessment tool, a Beach Vulnerability Index (BVI). Hourly wave and water level observations and daily beach profile data from 1993 to 2000 were collected at Hasaki Oceanographic Research Station (HORS), Japan and hourly tidal predictions were also used in the present study. Data-driven linear regression models were trialled on the eight-year data set for predicting storm-induced beach erosion. Shoreline change (dSL) and volume change (dV) were used as the main indicators of beach erosion. Wave energy flux, initial shoreline position, and maximum surge were selected as input variables for the regression models by using a correlation analysis. To quantify the shoreline change vulnerability (VI_dSL) and volume change vulnerability (VI_dV) to storm-induced erosion, five vulnerability classes were defined based on the first 5 years of shoreline and volume change data. BVIs were then calculated using VI_dSL and VI_dV in the testing period. During the study period, 48 storms were identified of which the first 38 were used to create two regression models. R² values of 0.58 and 0.52 were achieved for dSL and dV during the training period and during testing, they were 0.48 and 0.52 respectively. Majority of BVI predictions were slightly over-estimated, for a vulnerability study, this is preferable to having under-estimations. In conclusion, linear regression models provided moderate predictions of beach erosion measurements and adequate predictions of beach vulnerability.