3D Web Application of Wind Prediction Under Typhoon on Coastal Cities Based on Ensemble k-NN and OpenFOAM

Coastal cities are vulnerable areas to the impacts of strong wind under typhoon events. To anticipate the impact, this study proposes a novel approach to integrate the typhoon track predictions generated by an ensemble k-Nearest Neighbors (k-NN) as the reference locations of the typhoon center. The global weather forecast data corrected using an ensemble k-NN-based typhoon track was transferred to the OpenFOAM model simulation to obtain the urban-scale wind with multiple obstacles. The Gwangan Bridge, which faces the East Sea and is one of the sites in Busan City used in this study, is used for evaluating the approach with three typhoon events, i.e. Nari in 2007, Chaba in 2016, and Cimaron in 2018. The simulation results are disseminated in a 3D web application framework using Node JS for the web framework, PostgreSQL and PostGIS for database management, and CesiumJS for the main JavaScript library of 3D visualization. The study results show that this approach can performed well when compared with observed data from the wind station on the bridge, with correlation coefficient, mean absolute error (MAE), and root mean square error (RMSE) of 0.902, 1.505, and 1.927, respectively. The adjustment of grid positions through an ensemble k-NN-based typhoon prediction model on the global weather forecast data for OpenFOAM input significantly contributes to the accuracy of urban-scale wind prediction. This study also produces a more sophisticated web-based virtual globe and 3D city model to demonstrate a current typhoon prediction and past typhoons, online simulation results of ensemble k-NN algorithm, a volumetric profile of typhoons, wind flow, streamlines, wind pressure around buildings and a bridge, warning status, and time series of predicted wind velocity for supporting the disaster early warning system.