OmbriaNet—Supervised Flood Mapping via Convolutional Neural Networks Using Multitemporal Sentinel-1 and Sentinel-2 Data Fusion

Abstract

Regions around the world experience adverse climate change induced conditions which pose severe risks to the normal and sustainable operations of modern societies. Extreme weather events, such as floods, rising sea-levels and storms, stand as characteristic examples that impair the core services of the global ecosystem. Especially floods have a severe impact on human activities, hence early and accurate delineation of the disaster is of top-priority since it provides environmental, economic, and societal benefits and eases relief efforts. In this work, we introduce OmbriaNet, a deep neural network architecture, based on Convolutional Neural Networks (CNNs), that detects changes between permanent and flooded water areas by exploiting the temporal differences among flood events extracted by different sensors. To demonstrate the potential of the proposed approach, we generated OMBRIA, a bitemporal and multimodal satellite imagery dataset for image segmentation through supervised binary classification. It consists of a total number of 3.376 images, Synthetic Aperture Radar (SAR) imagery from Sentinel-1, and multispectral imagery from Sentinel-2, accompanied with ground truth binary images produced from data derived by experts and provided from the Emergency Management Service of the European Space Agency Copernicus Program. The dataset covers 23 flood events around the globe, from 2017 to 2021. We collected, co-registrated and pre-processed the data in Google Earth Engine. To validate the performance of our method, we performed different benchmarking experiments on the OMBRIA dataset and we compared with several competitive state-of-theart techniques. The experimental analysis demonstrated that the proposed formulation is able to produce high-quality flood maps, achieving a superior performance over the state-of-theart.We provide OMBRIA dataset, as well as OmbriaNet code at: https://github.com/geodrak/OMBRIA.

Segmentation and Visualization of Flooded Areas Through Sentinel-1 Images and U-Net

“…Due to this need, labeled image datasets have been used. Some datasets used in different proposals for flood analysis and mapping are Sen1Floods11 [23], which has Sentinel-1 and Sentinel-2 images of 11 manually labeled flood events; UNOSAT [24], with Sentinel1-SAR labeled images over 15 flood events; OMBRIA [25], with images labeled Sentinel-1 and Sentinel-2 over 23 floods; SEN12-FLOOD [26] with images labeled Sentinel-1 and Sentinel-2 and World Floods that contains information on 119 floods that occurred from 2015 to 2019. These datasets are used in different flood analysis proposals [21], [25], [27]- [29].…”

Mapping Invasive Aquatic Plants in Sentinel-2 Images Using Convolutional Neural Networks Trained With Spectral Indices

“…In [2], coastal ecosystem health status is evaluated. Other studies, such as [3] and [4], mapped flooded areas. In [5], polar sea ice leads in open water were mapped.…”

Cross-modal distillation for flood extent mapping

“…Although these methods have good performance, their utility at inference time is limited because of the cloud cover issues mentioned previously. Another line of work fuses Sentinel-1 and Sentinel-2 images (Tavus et al, 2020; Bai et al, 2021; Konapala et al, 2021; Drakonakis et al, 2022) to enhance surface water detection during flooded events. These methods not only require a cloud-free Sentinel-2 imaget also require that both images are acquired at about the same time to avoid alignment issues.…”