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.

Remote-sensing techniques offer an efficient alternative for mapping mining environments and assessing the impacts of mining activities. Airborne multispectral data in the thermal region and hyperspectral data in the optical region, acquired with the Airborne Hyperspectral Scanner (AHS) sensor over the Sokolov lignite open-pit mines in the Czech Republic, were analyzed. The emissivity spectrum was calculated for each vegetation-free land pixel in the longwave infrared (LWIR)-region image using the surface-emitted radiation, and the reflectance spectrum was derived from the visible, near-infrared and shortwave-infrared (VNIR–SWIR)-region image using the solar radiation reflected from the surface, after applying atmospheric correction. The combination of calculated emissivity, with the ability to detect quartz, and SWIR reflectance spectra, detecting phyllosilicates and kaolinite in particular, enabled estimating the content of the dominant minerals in the exposed surface. The difference between the emissivity values at λ = 9.68 µm and 8.77 µm was found to be a useful index for estimating the relative amount of quartz in each land pixel in the LWIR image. The absorption depth at around 2.2 µm in the reflectance spectra was used to estimate the relative amount of kaolinite in each land pixel in the SWIR image. The resulting maps of the spatial distribution of quartz and kaolinite were found to be in accordance with the geological nature and origin of the exposed surfaces and demonstrated the benefit of using data from both thermal and optical spectral regions to map the abundance of the major minerals around the mines.