In recent years, remote sensing analysis in the context of armed conflicts have gained increasing interest due to the capability of providing continuous and objective information over large areas. The destruction of the Nova Kakhovka dam and the subsequent discharge of the Kahkovske reservoir marked an significant event in the ongoing conflict in Ukraine with far-reaching consequences for water management, agriculture and energy planning. In this contribution, we reconstruct the drying out of the Kakhovske reservoir using Sentinel-1/2 time series data. We utilize existing, well-established methodologies to extract surface water information from optical and SAR satellite imagery and build multi-sensor time series. Water segmentation results are validated using very high resolution (VHR) WorldView-2 acquisitions.

Remote sensing data has become an essential part of today’s disaster management activities. In recent years, the German Aerospace Center (DLR) has developed a multi-sensor system for automatic and large-scale surface water extraction to support disaster management activities in the context of hydrological extreme events on a global level. The system consists of several cloud-based, automated processing chains that uses convolutional neural networks (CNN) to extract the surface water extent from radar and multi-spectral satellite data (e.g. Sentinel-1/2, Landsat, TerraSAR-X, very high-resolution (VHR) optical satellite missions). Within this contribution, four application examples of the proposed system in the frame of floods and hydrologic drought events are given.

Current research and development efforts at DLR`s Center for Satellite Based Crisis Information (ZKI) focus on deploying automated image analysis methods as part of rapid mapping processing routines. The use of machine learning methods enables processing of large amounts of heterogeneous satellite, aerial and drone images at varying spatial scales and temporal frequencies. In this work, we introduce an automated and scalable image processing chain for rapid building damage assessment, optimize it for inference on different hardware and provide application examples from recent natural disasters. We show the scalability of the method from high-frequency live-mapping with drones on a laptop to large-scale processing of satellite and aerial images on a high-performance computing cluster.

A key factor in successful flood mapping is the fast and easy access to water extent information at normal, non-flood conditions. In this study, we propose a fully automated and scalable methodology for generating and publishing information on permanent and seasonal surface water extent by leveraging cloud technologies such as Docker and Kubernetes. Products can be computed at global scale and made accessible free of cost via standardized interfaces. With this approach, the disaster response community benefits from easy data access and flexible integration into automated rapid mapping workflows. To showcase our approach, we generate reference water information in the scope of the 2022 flooding in Pakistan.

Recent advances in machine learning and the rise of new large-scale remote sensing datasets have opened new possibilities for automation of remote sensing data analysis that make it possible to cope with the growing data volume and complexity and the inherent spatio-temporal dynamics of disaster situations. In this work, we provide insights into machine learning methods developed by the German Aerospace Center (DLR) for rapid mapping activities and used to support disaster response efforts during the 2021 flood in Western Germany. These include specifically methods related to systematic flood monitoring from Sentinel-1 as well as road-network extraction, object detection and damage assessment from very high-resolution optical satellite and aerial images. We discuss aspects of data acquisition and present results that were used by first responders during the flood disaster.

The new, systematic Global Flood Monitoring (GFM) product of the Copernicus Emergency Management Service will provide a continuous monitoring of floods worldwide by immediately processing and analysing all incoming S-1 Interferometric Wide Swath data and making use of the data cube approach enabling a high product timeliness and the implementation of flood mapping algorithms that require data-driven model training. It integrates three independently developed flood mapping algorithms to improve the robustness and accuracy of the flood and water extent maps and to build a high degree of redundancy into the service.

In order to support disaster management activities related to flood situations, an automatic system for near-real time mapping of flood extent and duration using multi-sensor satellite data is developed. The system is based on four processing chains for the automatic derivation of the inundation extent from Sentinel-1 and TerraSAR-X radar as well as from optical Sentinel-2 and Landsat data. While the systematic acquisition plan of the Sentinel-1/2 and Landsat satellites allows a continuous monitoring of inundated areas at an interval of a few days, the TerraSAR-X processing chain has to be triggered on-demand over the disaster-affected areas. Beyond flood extent masks, flood duration products are generated to indicate the temporal stability and evolution of flood events. The flood monitoring system is demonstrated on a severe flood situation in Mozambique related to cyclone Idai in 2019.

Due to the similarity of the radar backscatter over open water and sand surfaces a reliable near real-time flood mapping based on radar sensors in arid areas is usually not possible. Within this paper an approach is presented to enhance the results of an automatic Sentinel-l flood processing chain of the German Aerospace Center (DLR) by removing overestimations of the water extent related to sand surfaces using a Sand Exclusion Layer (SEL) derived from time-series information of Sentinel-l data sets. The methodology was tested and validated on a flood event in May 2016 at Webi Shabelle River, Somalia, which has been covered by a time-series of 202 Sentinel-l scenes within the period April 2014 to May 2017. The algorithm proved capable to significantly improving the classification accuracy of the Sentinel-l flood service at this study site. Experimental results with variable lengths of the time-series have shown that the classification accuracy increased with increasing number of data sets at the cost of higher computational demand.

Within this paper activities of the International Charter `Space and Major Disasters', an international consortium of space agencies and satellite operators are described, and related contributions in terms of rapidly delivered satellite imagery and supporting services are presented. In particular, the role of the German Aerospace Center (DLR) is outlined, which became a member of the International Charter `Space and Major Disasters' in October 2010.

Fire is both a natural disturbance regime and a threat to infrastructure, forestry and human lives. Satellite remote sensing offers a fast and efficient way to reliably estimate the burnt area. In most cases, optical satellite data are used for burn scar detection. Nevertheless, smoke, clouds or rain can decrease the quality of classification. In these cases, SAR data can be a good alternative. Using quad-polarized SAR data, the backscatter can be decomposed into different scattering mechanisms, describing the scatterer more precisely. Here, we will present the possibilities and limitations of using quad-polarimetric ALOS-2 data for burnt area mapping using an object-based image analysis approach based on change-detection. As a case study the big fire event which affected Fort McMurray, Alberta, Canada in May/June 2016 was investigated.

Given the proven effectiveness of the split-based approach (SBA) for SAR image analysis in literature, the objective of this article focuses on designing a more efficient and robust version of the SBA for applications in the context of rapid flood mapping. A hierarchical tile-ranking SBA is proposed in this paper which is combined with a previous multilevel tile contrast analysis to significantly reduce the amount of data for the estimation of global threshold. A separability test is further applied to reject badly located tiles. The classification is optimized by merging pixel backscatter values, cluster size and local slope into a fuzzy-logic based post-classification framework. The proposed method was tested on Sentinel-1 SAR data acquired over Lake Liambezi in the Caprivi strip of Namibia and validated with respect to a Landsat-8 scene. Compared to tiles selected by the conventional SBA the proposed method automatically select better relevant ones and the classification is more robust with less misclassification of water-lookalikes.

This article presents a methodology to improve the flood classification results of an automatic Sentinel-1 Flood Service (S-1FS) in arid areas, where reliable SAR-based water detection is usually not possible. Statistical information of Sentinel-1 (S-1) backscatter time series data were used to remove water look-alikes related to sand surfaces which generally lead to significant overestimations of the flood extent as these are characterized by similar backscatter values as open water areas. The approach was demonstrated and evaluated on a flood event in May 2016 at Webi Shabelle River, Somalia, where a time series of 15 Sentinel-1 archive data were available. The algorithm proved capable to significantly improving the classification accuracy of the S-1 flood processing chain and increases the Overall Accuracy by approx. 5% and the User's Accuracy of the class flood by approx. 26% within the study area.

This article focuses on the mapping of fire burn scars, fire severity and soil erosion susceptibility using multi-sensoral satellite data. An automatic procedure for the mapping of fire affected areas and for the estimation of fire severity using Sentinel-2 data is presented. The Sentinel-2 based classification results are compared to a burn scar derived by a semi-automatic object-based approach using Sentinel-1 amplitude and coherence time-series data systematically processed by the Sentinel-1 InSAR Browse service implemented on the Geohazard Exploitation Platform (GEP) of ESA. Further, a transferable approach to compute a soil erosion susceptibility index based on Pléiades data is presented. The SAR- and optical-data based methods are applied in a test area near Marseille/Vitrolles, France, which was affected by severe forest fires in August 2016.

This article presents a semi-automated methodology for mapping of flooded areas with a special focus on flooded vegetation based on polarimetric Synthetic Aperture Radar (SAR) data. C-band SAR data is well suited for mapping of open water areas, while L-band enables the extraction of detailed information of flooded vegetation. Here, dual-pol C-band data of Sentinel-1 (S-1) is combined with quad-pol L-band ALOS-2/PALSAR-2 data to enable an accurate mapping of the entire flooded area. The developed procedure combines polarimetric decomposition based unsupervised Wishart classification with object-based post-classification refinement as well as the integration of spatial contextual information and global auxiliary data. The methodology was tested at the Evros River (Greek/Turkish border region), where a flooding event occurred in spring 2015.

Rapid Mapping is a mature Earth Observation (EO) service with many years of user oriented development since the International Charter `Space and Major Disaster' was established in 1999. This activity of providing EO satellite data derived disaster mapping during emergencies to civil protection and humanitarian user communities occurs at national, continental and worldwide scales. In an ESA R&D project the DLR and ICube-SERTIT are collaborating on enhancing automation of disaster mapping chains within increasingly fast production and monitoring cycles and Big Data handling considerations. The work uses optical and radar data from French, German, ESA Sentinel, and Copernicus Emergency Management Service rapid mapping activation sources. In this paper selected preliminary results are presented with SERTIT concentrating on CNES backed ORFEO Toolbox developments and the DLR notably presenting its Sentinel-1 automatic flood mapping chain. Finally, perspectives are given for this on-going initiative with their pertinence to the field of rapid mapping.

In this letter, a new approach is proposed to classify tsunami-induced building damage into multiple classes using pre- and post-event high-resolution radar (TerraSAR-X) data. Buildings affected by the 2011 Tohoku earthquake and tsunami were the focus in developing this method. In synthetic aperture radar (SAR) data, buildings exhibit high backscattering caused by double-bounce reflection and layover. However, if the buildings are completely washed away or structurally destroyed by the tsunami, then this high backscattering might be reduced, and the post-event SAR data will show a lower sigma nought value than the pre-event SAR data. To exploit these relationships, a rapid method for classifying tsunami-induced building damage into multiple classes was developed by analyzing the statistical relationship between the change ratios in areas with high backscattering and in areas with building damage. The method was developed for the affected city of Sendai, Japan, based on the decision tree application of a machine learning algorithm. The results provided an overall accuracy of 67.4% and a kappa statistic of 0.47. To validate its transferability, the method was applied to the town of Watari, and an overall accuracy of 58.7% and a kappa statistic of 0.38 were obtained.

This work presents a fully-automatic TerraSAR-X-based service for near real-time flood detection at a global level. Compared to semi-automatic flood detection approaches commonly applied in the rapid mapping community, automatic processing chains allow to reduce the critical time-span from the delivery of satellite data after flood events to the provision of satellite-derived crisis information (i.e. flood extent) to emergency management and decision makers. With respect to accuracy and computational effort, experiments performed on a data set of 150 different TerraSAR-X scenes acquired during flood situation all over the world with different sensor configurations confirm the robustness and effectiveness of the proposed flood mapping service. These results have been further confirmed by means of an in-depth validation performed for three study sites in Germany, Thailand, and Albania/Montenegro.

In this paper an automatic near-real time oil spill detection approach using single-polarized high resolution X-Band Synthetic Aperture Radar satellite data is presented. Dark formations on the water surface are classified in a completely unsupervised way using an automatic tile-based thresholding procedure. The derived global threshold value is used for the initialization of a hybrid multi-contextual Markov image model which integrates scale-dependent and spatial contextual information on irregular hierarchical graph structures into the segment-based labeling process of slick-covered and slick-free water surfaces. Experimental investigations performed on TerraSAR-X ScanSAR data acquired during large-scale oil pollutions in the Gulf of Mexico in May 2010 confirm the effectiveness of the proposed method with respect to accuracy and computational effort.

The near real-time provision of precise information about flood dynamics from synthetic aperture radar (SAR) data is an essential task in disaster management. A novel tile-based parametric thresholding approach under the generalized Gaussian assumption is applied on normalized change index data to automatically solve the three-class change detection problem in large-size images with small class a priori probabilities. The thresholding result is used for the initialization of a hybrid Markov model which integrates scale-dependent and spatiocontextual information into the labeling process by combining hierarchical with noncausal Markov image modeling. Hierarchical maximum a posteriori (HMAP) estimation using the Markov chains in scale, originally developed on quadtrees, is adapted to hierarchical irregular graphs. To reduce the computational effort of the iterative optimization process that is related to noncausal Markov models, a Markov random field (MRF) approach is defined, which is applied on a restricted region of the lowest level of the graph, selected according to the HMAP labeling result. The experiments that were performed on a bitemporal TerraSAR-X StripMap data set from South West England during and after a large-scale flooding in 2007 confirm the effectiveness of the proposed change detection method and show an increased classification accuracy of the hybrid MRF model in comparison to the sole application of the HMAP estimation. Additionally, the impact of the graph structure and the chosen model parameters on the labeling result as well as on the performance is discussed.

Medium resolution SAR satellite data have been widely used for water and flood mapping in recent years. Since the beginning of 2008 high resolution radar data with up to one meter pixel spacing of the TerraSAR-X satellite are operationally available. The improved ground resolution of the system offers a high potential for water detection. However, image analysis gets more challenging due to the large amount of image objects that are visible in the data. Water body detection methods are reviewed with regard to their applicability for TerraSAR-X data. Flood detection approaches for rapid disaster mapping are presented in this paper.