Developing near-real time flood mapping capabilities in Australia

Floods rank as the second-most deadly natural hazard in Australia, surpassed only by heatwaves. The ability to monitor flood extent and depth in near real-time is key to mitigating the loss of human life and minimizing the adverse socio-economic and environmental impacts. This study aims to discover the best way to map flood extent and depth in near-real time based on the most up-to-date  available information (i.e., gauge data, hydrological and hydrodynamic models, earth observations) in Australia. High resolution (i.e., 1-5 metres) airborne LiDAR DEMs are available across most of Australia's flood-prone east coast regions. The accessibility  of this information facilitates the creation of detailed, LiDAR-derived Height above Nearest Drainage (HAND) maps, which serve as an essential baseline for accurately mapping flood events. In gauged catchments, we utilized the Bureau of Meteorology’s environmental data management system, WISKI, an API solution that provides access to in-situ water levels at gauged locations across Australia. In ungauged catchments, we routed the Bureau’s operational runoff simulations (AWRA-L v7) using CaMa-flood to estimate flood level dynamics. By integrating these estimates into the HAND mapping approach, we generated a dynamic temporal profile of flood events in near-real time, effectively capturing the spatial-temporal onset, peak, and recession stages of flooding - essential information for emergency services. As the accuracy of the modelling approach is affected by uncertainties from runoff simulation and river morphology parameters, we additionally develop a multi-satellites-based flood monitor system to bolster the accuracy of modelled information. This system utilizes data from multiple medium-resolution satellite sources, including Sentinel-1 and -2, and Landsat -7 and -8/9. By extracting updated remote sensing imagery from Google Earth Engine and Digital Earth Australia, our approach simplifies and optimizes the process of deriving flood extent and depth from satellite and airborne LiDAR observations. Notably, this remote sensing approach significantly reduces interference from clouds, cloud shadows, terrain shadows, and vegetation cover, which are common challenges in optical remote sensing. Additionally, it effectively mitigates the 'double-bounce' effects often caused by vegetation and buildings in Synthetic Aperture Radar (SAR). To verify our end to end near real time flood mapping product, we used ICEYE (commercial SAR company) flood product to benchmark flood maps derived in this study and assessed the feasibilities of developing near-real time flood mapping network in Australia. Crucially, the immediate availability of data is essential in facilitating efficient allocation of resources and safeguarding infrastructure. Simultaneously, near real-time flood mapping plays a crucial role in enhancing community preparedness, allowing for strategic planning and swift action in response to hazardous situations.