Accurate flood delineation is crucial in many disaster management tasks, such as risk map production and update, impact estimation, claim verification, or planning of countermeasures for disaster risk reduction. Open remote sensing resources such as the data provided by the Copernicus ecosystem enable to carry out this activity, which benefits from frequent revisit times on a global scale. In the last decades, satellite imagery has been successfully applied to flood delineation problems, especially considering Synthetic Aperture Radar (SAR) signals. However, current remote mapping services rely on time-consuming manual or semi-automated approaches, requiring the intervention of domain experts. The implementation of accurate and scalable automated pipelines is hindered by the scarcity of large-scale annotated datasets. To address these issues, we propose MMFlood, a multimodal remote sensing dataset purposely designed for flood delineation. The dataset contains 1,748 Sentinel-1 acquisitions, comprising 95 flood events distributed across 42 countries. Along with satellite imagery, the dataset includes the Digital Elevation Model (DEM), hydrography maps, and flood delineation maps provided by Copernicus EMS, which is considered as ground truth. To provide baseline performances on the MMFlood test set, we conduct a number of experiments of the flood delineation task using state-of-art deep learning models, and we evaluate the performance gains of entropy-based sampling and multi-encoder architectures, which are respectively used to tackle two of the main challenges posed by MMFlood, namely the class unbalance and the multimodal setting. Lastly, we provide a future outlook on how to further improve the performance of the flood delineation task. Dataset and code can be found at https://github.com/edornd/mmflood.
INDEX TERMSComputer vision, deep learning, image processing, machine learning, semantic segmentation, remote sensing, natural disaster dataset. 24 ing the risk of severe flood events. The most recent reports 25 estimate that, by the end of this century, intense precipita-26 tion events that would typically occur two times per cen-27 tury would occur twice as often [1]. At the same time, 28 coastal areas are expected to see a constant sea-level rise 29 events: providing geographical extent, severity, and socioe-43 conomic impacts of a natural hazard using near real-time 44 Earth Observations (EO) can improve the assessment of the 45 affected areas more quickly and efficiently [6]. EO-based 46 imagery, either derived from aerial or satellite acquisitions, 47 is becoming crucial for the detection and the monitoring 48 of natural hazards, as well as to support activities related 49 to the restoration and adaptation phase. In recent years, the 50 increased accessibility of remote sensing services has allowed 51 for new and improved applications, from urban develop-52 ment [7] to agricultural [8], [9], and emergency scenar-53 ios [10], [11]. Considering floods, aerial or remote sensing 54 images have been extensively ...