dc.description.abstract | Synthetic Aperture Radar (SAR) imagery has been widely applied for flooding mapping based on change detection approaches. However, errors in the mapping result are expected since not all land-cover changes are flood-induced, and those changes are sensitive to SAR data, such as crop growth or harvest over agricultural lands, clearance of forested areas, and/or modifications to the urban landscape. This study, therefore, incorporated historical SAR images to boost the detection of flood-induced changes during extreme weather events, by applying a deep learning-driven spatiotemporal simulation framework, Convolutional Long Short-Term Memory (ConvLSTM), for simulating a synthetic image using Sentinel 1 intensity time series. This synthetic image can be prepared in advance of flood events, and then it can be used to detect flood areas using change detection when the post-image is available. Practically, a significant difference between the simulated and observed image is expected over inundated zones, which can be mapped by applying cut-off values to the Delta image (simulated minus observed image).
The proposed ConvLSTM-driven SAR-based change detection framework was applied to three different events from Isaac Region, Queensland Australia (Cyclone Debbie), Sofala Region, Mozambique (Cyclone Idai), and Belo Horizonte Metropolitan Area, Brazil (Brumadinho Dam Collapse). The generated Flood Proxy Maps were compared against reference data derived from Sentinel 2 and Planet Labs optical data. To corroborate the effectiveness of the proposed methods, Delta products were also generated for two baseline models (closest post-image minus pre-image and historical mean minus post-image) and two LSTM architectures: traditional LSTM and ConvLSTM. Results show that thresholding of ConvLSTM Delta yielded the highest Cohen’s Kappa coefficients in all study cases: 0.92 for Isaac Region, 0.78 for Sofala Province, and 0.68 for Belo Horizonte Metropolitan Area. Lower Kappa values obtained in the Mozambique case can be subject to the topographic effect on SAR imagery. These results still confirm the benefits in terms of classification accuracy that convolutional operations provide in time series analysis of Earth Observation satellite data employing spatially correlated information in a deep learning framework. | en_US |