Publication Date : 2016-11-22
Author : Garcia, F. C. C.Retamar, A. E.Javier, J. C.
Disaster Management Theme :
Disaster Type : Flood
Document Type : Research Paper
Languange : en
Link : http://ieeexplore.ieee.org/document/7848657/
DOST-Advanced Science and Technology Institute has installed various hydro-meteorological devices, such as Automated Rain Gauge(ARG), Water Level Monitoring Stations (WLMS), and Tandem Stations, all over the Philippines since 2010. While the stations provide valuable near real-time data for monitoring major riven basins, ahead-of-time flood estimations are of interest for early warning purposes especially for local communities situated along the river basin. This study addresses the need on developing a predictive model that can provide an ahead of time nowcasting system for water level and flood hazard to provide a decision support tool for the local communities. A data driven approach using machine learning is implemented to generate ahead-of-time water level estimation. Results from the testing data shows that the resulting model was able to provide an accurate ahead of time water level prediction without relying on rainfall-runoff models.