Satellite remote sensing is a powerful tool to map flooded areas. In recent years, the availability of free satellite data significantly increased in terms of type and frequency, allowing the production of flood maps at low cost around the world. In this work, we propose a semi-automatic method for flood mapping, based only on free satellite images and open-source software. The proposed methods are suitable to be applied by the community involved in flood hazard management, not necessarily experts in remote sensing processing. As case studies, we selected three flood events that recently occurred in Spain and Italy. Multispectral satellite data acquired by MODIS, Proba-V, Landsat, and Sentinel-2 and synthetic aperture radar (SAR) data collected by Sentinel-1 were used to detect flooded areas using different methodologies (e.g., Modified Normalized Difference Water Index, SAR backscattering variation, and supervised classification). Then, we improved and manually refined the automatic mapping using free ancillary data such as the digital elevation model-based water depth model and available ground truth data. We calculated flood detection performance (flood ratio) for the different datasets by comparing with flood maps made by official river authorities. The results show that it is necessary to consider different factors when selecting the best satellite data. Among these factors, the time of the satellite pass with respect to the flood peak is the most important. With co-flood multispectral images, more than 90% of the flooded area was detected in the 2015 Ebro flood (Spain) case study. With post-flood multispectral data, the flood ratio showed values under 50% a few weeks after the 2016 flood in Po and Tanaro plains (Italy), but it remained useful to map the inundated pattern. The SAR could detect flooding only at the co-flood stage, and the flood ratio showed values below 5% only a few days after the 2016 Po River inundation. Another result of the research was the creation of geomorphology-based inundation maps that matched up to 95% with official flood maps.

Accurate early warning systems for landslides are a reliable risk-reduction strategy that may significantly reduce fatalities and economic losses. Several machine learning methods have been examined for this purpose, underlying deep learning (DL) models’ remarkable prediction capabilities. The long short-term memory (LSTM) and gated recurrent unit (GRU) algorithms are the sole DL model studied in the extant comparisons. However, several other DL algorithms are suitable for time series forecasting tasks. In this paper, we assess, compare, and describe seven DL methods for forecasting future landslide displacement: multi-layer perception (MLP), LSTM, GRU, 1D convolutional neural network (1D CNN), 2xLSTM, bidirectional LSTM (bi-LSTM), and an architecture composed of 1D CNN and LSTM (Conv-LSTM). The investigation focuses on four landslides with different geographic locations, geological settings, time step dimensions, and measurement instruments. Two landslides are located in an artificial reservoir context, while the displacement of the other two is influenced just by rainfall. The results reveal that the MLP, GRU, and LSTM models can make reliable predictions in all four scenarios, while the Conv-LSTM model outperforms the others in the Baishuihe landslide, where the landslide is highly seasonal. No evident performance differences were found for landslides inside artificial reservoirs rather than outside. Furthermore, the research shows that MLP is better adapted to forecast the highest displacement peaks, while LSTM and GRU are better suited to model lower displacement peaks. We believe the findings of this research will serve as a precious aid when implementing a DL-based landslide early warning system (LEWS).