Landslides are one of the most widespread geohazards in Europe, producing significant social and economic impacts. Rapid population growth in urban areas throughout many countries in Europe and extreme climatic scenarios can considerably increase landslide risk in the near future. Variability exists between European countries in both the statutory treatment of landslide risk and the use of official assessment guidelines. This suggests that a European Landslides Directive that provides a common legal framework for dealing with landslides is necessary. With this long-term goal in mind, this work analyzes the landslide databases from the Geological Surveys of Europe focusing on their interoperability and completeness. The same landslide classification could be used for the 849,543 landslide records from the Geological Surveys, from which 36% are slides, 10% are falls, 20% are flows, 11% are complex slides, and 24% either remain unclassified or correspond to another typology. Most of them are mapped with the same symbol at a scale of 1:25,000 or greater, providing the necessary information to elaborate European-scale susceptibility maps for each landslide type. A landslide density map was produced for the available records from the Geological Surveys (LANDEN map) showing, for the first time, 210,544km 2 landslide-prone areas and 23,681 administrative areas where the Geological Surveys from Europe have recorded landslides. The comparison of this map with the European landslide susceptibility map (ELSUS 1000 v1) is successful for most of the territory (69.7%) showing certain variability between countries. This comparison also permitted the identification of 0.98Mkm 2 (28.9%) of landslide-susceptible areas without records from the Geological Surveys, which have been used to evaluate the landslide database completeness. The estimated completeness of the landslide databases (LDBs) from the Geological Surveys is 17%, varying between 1 and 55%. This variability is due to the different landslide strategies adopted by each country. In some of them, landslide mapping is systematic; others only record damaging landslides, whereas in others, landslide maps are only available for certain regions or local areas. Moreover, in most of the countries, LDBs from the Geological Surveys co-exist with others owned by a variety of public institutions producing LDBs at variable scales and formats. Hence, a greater coordination effort should be made by all the institutions working in landslide mapping to increase data integration and harmonization.

We propose a novel approach for rapid segmentation of flooded buildings by fusing multiresolution, multisensor, and multitemporal satellite imagery in a convolutional neural network. Our model significantly expedites the generation of satellite imagery-based flood maps, crucial for first responders and local authorities in the early stages of flood events. By incorporating multitemporal satellite imagery, our model allows for rapid and accurate post-disaster damage assessment and can be used by governments to better coordinate medium-and long-term financial assistance programs for affected areas. The network consists of multiple streams of encoder-decoder architectures that extract spatiotemporal information from medium-resolution images and spatial information from high-resolution images before fusing the resulting representations into a single medium-resolution segmentation map of flooded buildings. We compare our model to state-of-the-art methods for building footprint segmentation as well as to alternative fusion approaches for the segmentation of flooded buildings and find that our model performs best on both tasks. We also demonstrate that our model produces highly accurate segmentation maps of flooded buildings using only publicly available medium-resolution data instead of significantly more detailed but sparsely available very high-resolution data. We release the first open-source dataset of fully preprocessed and labeled multiresolution, multispectral, and multitemporal satellite images of disaster sites along with our source code.

No abstract

No abstract