Canada's Emergency Geomatics Service Near Real-Time Flood Mapping from Multi-Source Data

Abstract

The Canadian Emergency Geomatics Service (EGS) creates and disseminates flood maps in near real time during major flood events. In 2017, the EGS developed and deployed a fully automated method to map open water and flooded vegetation using available optical and radar imagery. This method employs machine learning trained using historical inundation maps to classify open water followed by region growing to map flooded vegetation, and requires only a few sensor-specific parameters for different input satellite data. Recent collection of High-Resolution Digital Elevation Model (HRDEM) lidar data over important floodplains, in Canada, in combination with in-situ flood perimeter observations from an EGS-developed Citizen Geographic Information (CGI) mobile application or other sources, has facilitated the development of better urban flood mapping where traditional satellite-based methods fail. This presentation will describe existing operational EGS flood mapping methods with examples from previous activations, as well as new developments in urban flood mapping that will become operational depending on the availability ofrequired input data.