Zhouyayan Li, Ibrahim Demir

Remote Sensing (RS) imagery is an important data source in surface water mapping applications thanks to its high spatial and temporal consistency and scalability. The introduction of Google Earth Engine (GEE) has cleared some of the major barriers of fast and large-scale RS-based geospatial analyses by providing easy and open access to most of the commonly used RS image products as well as built-in functions designed for geospatial analysis. There is a growing interest in developing GEE applications that can work for different regions and time durations to improve the reusability of GEE scripts and reduce manual effort during the entire workflow of water-body extraction. Despite all those advancements and efforts, there is still a need for creating GEE applications that are user-friendly and can serve both remote sensing experts and students. These applications are also expected to be powerful and comprehensive enough to handle each step along the entire lifecycle of water body extraction workflow and are capable of handling geomorphic and geospatial discrepancies between regions under various configurations. Given these needs and challenges, this study presents the MultiRS Flood Mapper, a GEE application that incorporates three most used RS imagery (i.e., Sentinel-1 SAR, Landsat 8, and Sentinel-2) in water body mapping and integrates advanced dynamic thresholding algorithms and powerful postprocessing modules to improve classification results under the influence of dense vegetation and cloud, and in regions with constrained hydraulic conditions. In addition, the MultiRS Flood Mapper comes with a self-explanatory and user-friendly interface. Most functional modules for RS image processing that require professional knowledge are fully automated and the remaining function in an intuitive and interactive way, which therefore enables the MultiRS Flood Mapper to have great potential to serve a broad audience with various backgrounds and purposes.